The 1 repetition maximum (1RM) technique is widely accepted as the most valid measurement of dynamic strength and has been used extensively as the primary evaluation of the success of resistance training programs at all levels. However, many college football teams have begun to adopt the NFL-225 test as their primary measure of upper-body strength, perhaps because of its use at the Nation Football League testing combine (23). The NFL-225 test is a muscular endurance test in which a player attempts to perform as many repetitions as possible with an absolute weight of 225 lbs. Previous studies have produced high correlations (r > 0.91) between the number of repetitions-to-fatigue (RTF) with an absolute load and the 1RM (3, 20, 21, 26, 32). Collectively, these studies indicate that individuals with greater strength can produce more RTF before fatigue causes muscle failure.
A major problem identified with the NFL-225 test is the increasing error of prediction when the RTF exceed 10 (20, 21). Although the error when using less than 10 RTF is in the acceptable range of 11 lbs, it increases to as much as 17 lbs when the repetitions exceed 10 (21). Whisenant et al. (32) recently noted that the addition of ethnicity improved prediction accuracy when added to NFL-225 regression equations. However, in developing their own prediction equation, they did not use ethnicity as a major contributing variable. Instead, lean body mass (LBM), height, and RTF were the only variables included in the prediction equation.
Previous studies have shown the potential of using anthropometric dimensions to predict 1RM strength in the bench press (8, 10, 16–19). Flexed arm circumference, or circumference corrected for triceps skinfold, and chest circumference have commonly been selected by regression analysis as major predictors of 1RM performance. Additional items selected in athletic populations have included arm length and/or drop distance (i.e., the distance the bar travels from full-arm extension to touching the chest). Although combinations of these variables have produced reasonably high correlations with the 1RM (r = 0.71 to 0.83; Refs. 8, 16–19), the standard errors of estimate have been larger than desirable for precise prediction of 1RM (SEE = 25 to 30 lbs).
Only a few studies have attempted to combine the 2 approaches of anthropometric dimensions and muscular endurance performance in an effort to increase the precision of predicting bench press strength. Cummings and Finn (4) found that shoulder width made an additional contribution to repetition weight (RepWt) and RTF in the prediction of 1RM bench for untrained women. Recently, Kravitz et al. (10) noted that several anthropometric dimensions did little to enhance the predictive accuracy above that of RTF and RepWt for estimating the 3 powerlifts in elite adolescent lifters.
Due to the growing popularity of the NFL-225 test as a means of predicting the upper-body strength of college football players, it would be beneficial to determine if simple anthropometric dimensions could reduce the error associated with prediction from RTF. The additional method, however, would need to be quick, easy, and simple in order to be time-efficient in the evaluation process of players. Therefore, the purpose of this study was to determine the degree to which the addition of selected anthropometric dimensions to RTF performance using 225 lbs would decrease the error of predicting 1RM bench press in college football players.
Experimental Approach to the Problem
This study sought to determine the potential of anthropometric dimensions to reduce the prediction error associated with using an absolute muscular endurance task such as the NFL-225 test to estimate 1RM bench press performance in college football players. As part of their routine off-season training, all participants performed a 1RM and as many repetitions as possible using a weight of 225 lbs. In addition, selected physical characteristics were measured by an experienced investigator. These measurements included skinfolds, muscle circumferences, and skeletal lengths.
Sixty-one players from an NCAA Division II college football team were evaluated after giving their informed consent to participate. The study was approved by the Institutional Review Board for Human Subjects. Each player had undergone a minimum of 8 weeks of heavy resistance training during the winter off-season conditioning program prior to measurement.
The off-season program focused on low repetitions and heavy loads and emphasized a periodized methodology for core exercises such as the bench press, squats, deadlifts, and push presses. In addition, supplemental exercises such as incline presses, lat pulls, and upright or bent-over rowing were employed. Players were measured the week following the last workout of the cycle to allow sufficient recovery to achieve peak performance. The highly competitive atmosphere of the team testing environment provided sufficient motivation to promote superior performance from each player.
1RM Test and NFL-225 Test
The 1RM bench press procedure followed a standard “touch-and-go” protocol in which the bar was required to be lowered slowly to touch the chest before being pressed immediately to full arms’ extension (27). During testing, each player was allowed to warm up according to personal preferences using light weights of approximately 50–75% of estimated 1RM. When testing began, a weight was selected by the player, and 1 repetition was performed. Following this repetition (and all others), a minimum of 5 minutes rest was given (30) before attempting subsequent repetitions with additional weight. Most players reached their 1RM within 3–5 attempts. Standard Olympic bars and plates were used for all lifts, and the player used a grip that was slightly wider (approximately 15–35 cm) than shoulder width (28).
During the week following the 1RM testing, each player performed as many repetitions as possible using a load of 225 lbs. Following individual warm-ups, the player grasped the bar at the same position used during the 1RM procedure. Although no cadence was set for the repetition tests (11), each player was encouraged to maintain a constant pace but was allowed no more than a 2-second pause between each repetition. The bar was required to touch the chest on each repetition, but the subject was admonished not to bounce the bar off the chest and to return it to full-arm extension on each repetition. The head, upper back, and buttocks were required to remain in contact with the bench throughout the test. The test was terminated when the subject could not complete a repetition with proper form.
Anthropometric dimensions included measurements of flexed dominant arm circumference, midexpiration chest circumference, arm length, drop distance, and five skinfolds. All measurements were taken in triplicate and averaged to represent a site. Flexed arm circumference was measured with a cloth tape around the maximum girth of the arm (13). Chest circumference was measured at the level of the nipples at midexpiration with the arms held at the sides (13). Arm length was measured from the acromion process to the olecranon process using a broad-blade anthropometer (13). Drop distance was measured from the bar to the sternum with the subject in the supine position on a standard bench press bench holding an empty Olympic bar. Skinfold measurements were made at the triceps, subscapular, suprailiac, abdominal, and thigh sites using Harpenden calipers (13).
Derived anthropometric dimensions included arm cross-sectional area (CSA), percent fat (%fat), and LBM. Arm CSA was calculated according to the de Koning et al. (5) method of differential skinfold correction of flexed arm circumference. Body density was estimated using the Lohman equation (12) and converted to %fat using the Siri equation (25).
In addition to using current NFL-225 prediction tests (Table 1), other widely accepted RTF equations (Table 2) and anthropometric prediction equations were evaluated (Table 3). Although many of these equations do not have a statistically sound basis to their development, they are widely used nevertheless to estimate the 1RM. The addition of anthropometric dimensions to the NFL-225 prediction equations was also evaluated to determine if prediction error could be reduced.
Mean and SD were calculated for each variable. Pearson correlation was used to determine the relationship of selected variables with 1RM bench press. The potential of previously reported repetition prediction equations (Table 1 and 2) and anthropometric prediction equations (Table 3) to estimate 1RM was evaluated using Pearson correlations and paired t-tests. Multiple linear regression was used to evaluate the predictive potential of combining RTF with anthropometric variables for estimating 1RM bench press in the current sample.
Physical and performance characteristics of the players are shown in Table 4. The highest correlation with 1RM bench press was for RTF using 225 lbs, accounting for 92% of the explained variance. Muscle circumference variables accounted for approximately 42–48% of the explained variance in 1RM bench press, which dropped to 9–21% when the effect of body mass was removed. The explained variance in 1RM from RTF decreased only to 83% when arm and chest circumferences were removed from the relationship.
Of the previously developed NFL-225 prediction equations, all produced high correlations between the predicted and actual 1RM values (Table 5). However, only the Chapman et al. (3) equation produced predicted values that were not significantly different from actual 1RM values. The Allenheilgen and Chapman et al. (3) equations predicted over 60% of the players in the current sample within ±10 lbs of their actual 1RM performance. Each of these equations was constructed using a single team. The Mayhew et al. equations (20, 21), developed using composite samples of various competitive levels of players, were less successful in the number of players that were predicted within ±10 lbs of their actual 1RM performance. The Whisenant et al. (32) equation was the least successful at approximating 1RM performance.
Generalized curvilinear equations that have previously been identified for use when >10 RTF are completed were moderately successful in predicting 1RM (Table 6). Only the Mayhew et al. equation (15) produced predicted values that were nonsignificantly different from actual 1RM performances. It was also the only equation to produce more than 50% of the players with predicted values within ±10 lbs of their actual 1RM performance. Combining the Mayhew et al. RTF equation (15) with the Chapman et al. NFL-225 equation (i.e., averaging the 2 values) produced nonsignificant underestimations of 1RM performance (-0.9 ± 14.9 lbs) with a correlation coefficient equivalent to either method taken singularly (r = 0.96). This approach was slightly less effective in predicting the number of players within ± 10 lbs of the actual 1RM (57.3%) than was using RTF alone.
Four anthropometric prediction equations used to estimate 1RM bench press produced significant but only moderately high correlation coefficients between predicted and actual 1RM values (Table 7). An equation developed on average college men (16) and 1 developed on football players (18) were the only ones that produced nonsignificant differences between predicted and actual 1RM values. None of the equations estimated more than 50% of the players within ± 10 lbs of their actual 1RM performance. Combining the anthropometric football equation (18) with the Chapman et al. (3) NFL-225 equation (i.e., averaging the 2 predictions) produced a nonsignificant overestimation of 1RM performance (3.8 ± 20.6 lbs) with a correlation coefficient slightly lower than either method taken singularly (r = 0.89). This approach produced only 36.1% of the predicted values within ±10 lbs of the actual 1RM.
In order to explore possible explanations for over- or underpredictions, ratios were constructed by dividing actual bench press performance by predicted performance from both the anthropometric football-2 equation (19) and the Chapman et al. (3) NFL-225 equation. The ratio derived from the RTF prediction had a smaller variation (Actual 1RM/Predicted 1RM = 1.00 ± 0.04) than the ratio derived from anthropometric prediction (Actual 1RM/Predicted 1RM = 0.98 ± 0.12). The distribution of each ratio was categorized into the top, middle, and bottom thirds, and players were classified as overachievers, average achievers, and underachievers, respectively. A performer classification was then developed by comparing achiever categories. If the achiever categories for each ratio were equivalent, the player was labeled an average performer. If the achiever category for the RTF ratio was better than for the anthropometric ratio classification, the player was labeled a high performer. If the achiever category for the anthropometric ratio was better than for the RTF ratio classification, the player was labeled a low performer. One-way analyses of variance (ANOVAs) for anthropometric variables indicated that low performers were significantly greater in body size (body mass and LBM) but had significantly lower muscle quality (1RM/CSA) than high performers (Table 8).
In order to evaluate the contribution of anthropometric dimensions to the prediction of 1RM bench press in the current sample, step-wise multiple regression analysis was employed. The RTF was the only significant variable selected to predict 1RM performance (r = 0.96, SEE = 12.3 lbs). The slope and intercept of the equation produced were similar to those of other NFL-225 equations:
If selected anthropometric dimensions were forced into the regression equation, none of them made more than a 1% contribution to the explained variance and did not reduced the SEE.
This study demonstrated that anthropometric dimensions do not improve the accuracy of predicting 1RM bench press strength from muscular endurance repetitions performed with an absolute load of 225 lbs in college football players. The greater error in prediction at the higher RTF levels could not be reduced with either estimates of local muscular size (i.e., arm and chest measurements) or overall body size (i.e., body mass or LBM). As in previous studies (20, 21), the prediction error was better when ≤10 RTF were performed (±8.9 lbs) than when >10 RTF were performed (±14.4 lbs). Over 92% of the variance in 1RM performance was accounted for by the RTF with the absolute load of 225 lbs, which could not be further enhanced by adding any combination of anthropometric dimensions. Whisenant et al. (32) recently indicated that ethnicity and selected anthropometric variables increased the variance accounted for by 0.2–1.1% for NFL-225 prediction equations. However, when they produced a multiple regression equation on their sample, it did not include ethnicity as a contributing variable for estimating 1RM performance. Their equation did include an estimate of LBM from a skinfold prediction equation, a technique that could take additional time to perform. Furthermore, the use of the Whisenant et al. (32) equation on our subjects accounted for only 88% of the variance in 1RM performance, which was approximately 4% lower than the other RTF prediction equations. If time efficiency is 1 of the considerations in the football testing program, adding body composition analysis to enhance strength prediction accuracy appears to be of little value.
Previously developed anthropometric strength prediction equations tended to overpredict 1RM performance by 2.9–11.8% and were not successful in accurately estimating actual strength within the narrow limit of ±10 lbs (Table 7). A football-specific equation (19) performed best but could produce only 34% of the players with accurate estimates (within ±10 lbs) of 1RM performance. When this anthropometric equation was combined with the Chapman et al. (3) NFL-225 equation, the number of players with predictions within ±10 lbs (36%) was less than when the Chapman et al. (3) equation alone was used. Sale et al. (24) have indicated that increases in muscle hypertrophy do not necessarily elicit increases in maximal strength. It is possible that the multiset programs used in resistance training for football may enhance muscle size to a greater degree than it develops strength in some players.
The question of prediction accuracy still remains. Most of the NFL-225 equations overestimated 1RM by an average of only 1.2–3.0%. In fact, all of the NFL-225 equations, except the Whisenant et al. (32) equation, produced more than half of the predicted values within ±3.3% of the actual performance values. This could be considered acceptable accuracy in light of the 4.4% variation in repeated trials for the 1RM (2). However, it was at the upper end of the strength range that prediction errors became greatest. Above 350 lbs (n = 15), the Chapman et al. (3) equation significantly underpredicted by an average of -11.3 lbs (±16.7 lbs), equivalent to -2.9% (±4.3%).
Noel et al. (23) have indicated that body weight increases in college football players at the Division I level over the past 2 decades can be attributable more to gains in fat mass than in lean body mass. They further indicated that there is likely a point of diminishing returns where performance begins to be compromised by the weight gain. The current study agrees in part with their conclusions in that the poorest performers relative to their predicted capabilities were also the heaviest individuals who had the highest %fat values (Table 7). However, where the current results on Division II players deviated from those on Division I players (23) was the change in body composition over recent years. When players in the current study were compared with a sample of Division II players from a decade ago (18), a 12% greater body mass was accomplished by a 12% gain in LBM and a 20% gain in fat weight. This resulted in predicted %fat values remaining practically the same between the previous and current samples (15.4% vs. 15.6%, respectively). Bench press strength increased by 8% during that period, which indicated that although Division II players have gained in muscle mass over the last decade, their gains in strength still seem to be lagging behind. Overall size at most playing positions remains the dominant issue at all levels in football today, rather than an optimal lean-to-fat ratio (23) or strength-to-muscle mass ratio.
One possible explanation of the difference in the predictive accuracy of muscular endurance repetitions to estimate maximal strength could be variations in muscle fiber type. Highly trained powerlifters have been shown to have a higher quality of muscle than untrained controls (6). The persistent use of heavy loads in multiple sets could shift muscle fiber types more toward the type IIa classification (6). Fry et al. (6) indicated that success in maximal force lifts such as the bench press may be influenced by a high percentage of these muscle fibers. It may well be that football players with a greater proportion of type I fibers are able to perform RTF with greater proficiency than they are able to produce single-effort, high-force generation. Conversely, players with a higher proportion of type II fibers may have a greater capability to develop maximal force than to produce endurance repetitions. Although differences in fiber type distribution could account for some of the over-and underpredictors when using RTF to estimate 1RM performance, there is not a simple approach to identifying this problem, and no practical indicator of fiber type is yet available.
Yet another possible explanation of the discrepancy between muscle endurance and muscle strength could lie in the neural activation process. Bamman et al. (1) pointed out that if neural activation is not 100% in all individuals, the differences in muscle size will not fully explain the differences in strength. In their study, a uniarticular muscle was measured during isometric contraction and produced a correlation between estimated flexor CSA and maximal voluntary contraction (r = 0.733) similar to that noted in the current study (r = 0.69). Therefore, it is possible that some of the players were not able to maximally activate muscles involved in the bench press either to their full potential or in the proper sequence.
Further examination of the neural adaptation that accompanies heavy resistance training was recently made by Judge et al. (9). They found that a periodized strength program produced a maximization of the neural drive during a single rapid isometric contraction. This would indicate that highly trained muscle may have greater motor unit synchronization for short, high-force contractions (7). However, Judge et al. (9) noted no increase in neural activation during a sustained 2-second maximal isometric contraction. Although their sustained contraction was much shorter than the average time of 28.5 seconds required to complete the NFL-225 tests (26), it points to the possibility that a lack of improvement in the asynchronous activation of motor units during a muscular endurance exercise compared with an enhanced synchronization during maximal effort could reduce the correlation between strength and endurance.
Whenever performance prediction is attempted there will always be some degree of error associated with the estimates. The fact that the standard deviation for the constant error in the current study was similar to the total error (a method that combines the degree of association between 2 variables with their degree of difference) for 5 of the 6 NFL-225 equations (Table 4) offers support for the potential of predicting 1RM performance from RTF in the majority of football players. In the 5 best equations (excluding the Whisenant et al.  equation), the agreement in prediction to within ±10 lbs was consistent in 41% of the players. In other words, 25 of the 61 players had predicted 1RM values within ± 10 lbs of their actual 1RM performance on all 5 equations. In the 18 players who did not have any prediction equations estimate that agreed with their actual performance within ±10 lbs, no clear-cut single determinant or combination of determinants could be established to explain their over- or underprediction. Therefore, the search for methods that can produce more accurate predictions at the higher end of the strength continuum must continue.
Strength is a vital part of modern football, and no team can expect to be successful without a well-planned, heavy resistance training program as an adjunct to the on-the-field skill work. In an effort to expedite the testing of muscle strength, several approaches have been investigated. The most feasible in terms of time efficiency and predictive accuracy appears to be the use of muscular endurance repetitions to predict 1RM. To that end numerous equations have been developed to estimate 1RM strength from the number of repetitions completed before muscular fatigue halts contraction. The current study indicates that the NFL-225 test can be used with reasonable accuracy to estimate the 1RM strength of the majority of players on a team. Although some error in prediction may occur, especially at the higher end of the repetition continuum, no simple anthropometric dimensions appear available to reduce that error. In addition, no simple technique is currently available to identify players who are significantly over- or under-estimated by prediction techniques. It may still be left to the strength and conditioning specialist to modify training programs based on the use of predicted 1RM values to achieve peak performance in individual players.
1. Bamman, M.M., B.R. Newcomer, D.E. Larson-Meyer, R.L. Weinsier, and G.R. Hunter. Evaluation of the strength-size relationship in vivo using various muscle size indices. Med. Sci. Sports Exerc.
2. Braith, R.W., J.E. Graves, S.H. Leggett, and M.L. Pollock. Effect of training on the relationship between maximal and submaximal strength. Med. Sci. Sports Exerc.
3. Chapman, P.P., J.R. Whitehead, and R.H. Binkert. The 225-lb reps-to-fatigue test as a submaximal estimate of 1-RM bench press
performance in college football players. J. Strength Cond. Res.
4. Cummings, B., and K.J. Finn. Estimation of a one repetition maximum bench press
for untrained women. J. Strength Cond. Res.
5. de Koning, F.L., R.A. Blinkhorst, J.M.G. Kauer, and H.O.M. Thijssen. Accuracy of an anthropometric estimate of the muscle and bone area in a transversal cross-section of the arm. Int. J. Sports Med.
6. Fry, A.C., J.M. Webber, L.W. Weiss, M.P. Harber, M. Vaczi, and N.A. Pattison. Muscle fiber characteristics of competitive power lifters. J. Strength Cond. Res.
7. Hakkinen, K. Neuromuscular and hormonal adaptations during strength and power training: a review. J. Sports Med. Phys. Fitness.
8. Hart, C.L., T.W. Ward, and J.L. Mayhew. Anthropometric correlates of bench press
performance following resistance training. Sports Train. Med. Rehabil.
9. Judge, L.W., C. Moreau, and J.R. Burke. Neural adaptations with sport-specific resistance training in highly skilled athletes. J. Sports Sci.
10. Kravitz, L., C. Akalan, K. Nowicki, and S.J. Kinzey. Prediction of 1 repetition maximum in high-school power lifters. J. Strength Cond. Res.
11. LaChance, P.F., and T. Hortobagyi. Influence of cadence on muscular performance during push-up and pull-up exercises. J. Strength Cond. Res.
12. Lohman, T.G. Skinfolds and body density and their relation to body fatness: a review. Hum. Biol.
13. Lohman, T.G., A.F. Roche, and R. Martorell (Eds.). Anthropometric Standardization Reference Manual.
Champaign, IL: Human Kinetics, 1988.
14. Lombardi, V.P. Beginning Weight Training.
Dubuque, IA: W.C. Brown, 1989.
15. Mayhew, J.L., T.E. Ball, M.D. Arnold, and J.C. Bowen. Relative Muscular endurance performance as a predictor of bench press
strength in college men and women. J. Appl. Sports Sci. Res.
16. Mayhew, J.L., T.E. Ball, T.E. Ward, C.L. Hart, and M.D. Arnold. Relationships of structural dimensions to bench press
strength in college males. J. Sports Med. Phys. Fitness.
17. Mayhew, J.L., T.P. McCormick, F.C. Piper, A.L. Kurth, and M.D. Arnold. Relationships of body dimensions to strength performance in novice adolescent male powerlifters. Pediatr. Exerc. Sci.
18. Mayhew, J.L., F.C. Piper, and J.S. Ware. Anthropometric correlates with strength performance among resistance trained athletes. J. Sports Med. Phys. Fitness.
19. Mayhew, J.L., F.C. Piper, and J.S. Ware. Relationships of anthropometric dimensions to strength performance in resistance trained athletes. J. Phys. Educ. Sport Sci.
20. Mayhew, J.L., J.S. Ware, M.G. Bemben, B. Wilt, T.E. Ward, B. Farris, J. Juraszek, and J.P. Slovak. The NFL-225 test as a measure of bench press
strength in college football players. J. Strength Cond. Res.
21. Mayhew, J.L., J.S. Ware, K. Cannon, S. Corbett, P.P. Chapman, M.G. Bemben, T.E. Ward, B. Farris, J. Juraszek, and J.P. Slovak. Validation of the NFL-225 test for predicting 1-RM bench press
performance in college football players. J. Sports Med. Phys. Fitness.
22. McGee, K.J., and L.N. Burkett. The National Football League combine: a reliable predictor of draft status? J. Strength Cond. Res.
23. Noel, M.B., J.L. VanHeest, P. Zaneteas, and C.D. Rodgers. Body composition in Division I football players. J. Strength Cond. Res.
24. Sale, D.G., J.E. Martin, and D.E. Moroz. Hypertrophy without increased isometric strength after weight training. Eur. J. Appl. Physiol.
25. Siri, W.E. Body composition from fluid spaces and density: analysis of methods. In: Techniques for Measuring Body Composition.
J. Brozek, ed. Washington, DC: National Academy of Science, 1961.
26. Slovak, J.P., T.E. Ward, and J.L. Mayhew. Cross-validation of NFL-225 tests for predicting 1 repetition maximum bench press
performance for a college football team [Abstract]. J. Strength Cond. Res.
27. Stone, M., and H. O'Bryant. Weight Training: A Scientific Approach.
Minneapolis, MN: Burgess, 1987.
28. Wagner, L.L., S.A. Evans, J.P. Weir, T.J. Housh, and G.O. Johnson. The effect of grip width on bench press
performance. Int. J. Biomech.
29. Wathen, D. Load assignment. In: Essentials of Strength Training and Conditioning.
T.R. Baechle, ed. Champaign, IL: Human Kinetics, 1994. pp. 435-439.
30. Weir, J.P., L.L. Wagner, and T.J. Housh. The effect of rest interval length on repeated maximal bench presses. J. Strength Cond. Res.
31. Welday, J. Should you check for strength with periodic max lifts? Scholas. Coach.
32. Whisenant, M.J., L.B. Panton, W.B. East, and C.E. Broeder. Validation of submaximal prediction equations for the 1 repetition maximum bench press
test on a group of collegiate football players. J. Strength Cond. Res.