Secondary Logo

Journal Logo


Designing Exercise Regimens to Increase Bone Strength

Turner, Charles H.1; Robling, Alexander G.2

Author Information
Exercise and Sport Sciences Reviews: January 2003 - Volume 31 - Issue 1 - p 45-50
  • Free



Regular exercise has effects on bone density, size, and shape, resulting in substantial improvements in mechanical strength. For example, the humeri of professional tennis players exhibit approximately 40% more cortical bone on the playing side (arm that holds the racquet) compared with the nonplaying side. Another striking example of how exercise affects bone properties emerged when a group of collegiate gymnasts were followed for 2 yrs, which captured an 8-month competitive season (vigorous loading) and a 4-month off-season (reduced loading) for each year. There was a clear increase in hip and spine bone mineral density (BMD) during the competitive seasons, followed by a decline in BMD during the off-seasons (12). The positive association between exercise and bone mass has prompted many physicians and public health officials to recommend that individuals engage in daily exercise, with the goal of reducing the incidence of osteoporotic fracture, and the morbidity/mortality that ensues. However, there is no clear consensus on exactly how one should exercise to reap the greatest returns in terms of bone health. What exercises are best? How often should one exercise? Is it sometimes better not to exercise?

Over the past several years, a number of studies from both the clinic and laboratory have been conducted to address strategies for maximizing the osteogenic effects of exercise. Below, we review some of the more pertinent outcomes of these investigations. Specifically, we address the bone cellular environment produced by different types of loading (exercise), the tendency of bone cells to ignore mechanical signals when exercise duration is prolonged, and the reestablishment of bone’s sensitivity to loading. Furthermore, we develop a new measure of effectiveness for exercise protocols called the “osteogenic index” (OI), which incorporates several recently discovered biological phenomena and allows one to reasonably predict the outcome of an exercise protocol on bone mass. Finally, we outline exercise recommendations for reducing the risk of osteoporotic fracture.


Proper exercise can add new bone and/or reduce bone loss ultimately to affect bone mass, but bone mass (or areal BMD (aBMD)) is merely a surrogate measure for bone strength. The real issue at hand is whether or not an individual will fracture the hip, spine, or wrist. Bone mineral content (BMC) and BMD are related to bone strength, but sometimes inferring strength from bone-mineral measurements can be misleading. We recently conducted an experiment on rats in which mechanical loading was applied to the right ulna of adult female rats three times · w−1 for 16 wk (7). The whole ulnae were scanned for BMC and areal BMD using dual energy x-ray absorptiometry (DXA), then broken in axial compression using a materials-testing machine. The load-induced increase in aBMD and BMC were modest, reaching 5.4% and 6.9%, respectively. Despite these small gains in bone mineral, mechanical testing revealed a 64% increase in ultimate force (the maximum amount of force the bone could support before failing) and a 94% increase in energy to failure (the amount of energy absorbed by the bone before failure).

The reason that a small amount of new bone resulted in such dramatic changes in bone strength is because the new bone formation was localized to the medial and lateral periosteal surfaces where mechanical strains (stresses) were greatest. Application of strain gauges around the circumference of the ulna midshaft demonstrated that the largest peak strain (stress) was in the medial quadrant. This region was loaded in compression. The lateral surface also had substantial peak strain, but in this region, the strain was tensile. The cranial and caudal surfaces had very low strain. Loading induced osteogenesis at the medial and lateral quadrants of the mid-ulna, with very little new bone added on the cranial and caudal quadrants (Fig. 1). Consequently, only modest increases in total new bone formation produced a large increase in bone strength by placing bone where the biomechanical demands were greatest.

Figure 1
Figure 1:
Adult female rats subjected to 16 wk of axial loading of the right ulna (360 cycles·d−1, 3 d·wk−1) show substantial increases in bone formation and strength but only modest increases in areal density and mass. (A) Darkfield photomicrographs of the midshaft ulna from an animal labeled with a fluorochrome 2 wk (Label 1) and 16 wk (Label 2) after loading began. Note that most of the new bone formation is located in the medial and lateral quadrants, where bending strains resulting from axial loading were greatest. (B) Dual energy x-ray absorptiometric (DXA) scans of whole ulnae revealed small gains (5–7%) in BMC and areal BMD (aBMD) as a result of loading. Mechanical testing of the same bones revealed substantial (64–94%) increases in ultimate force (FU) and energy to failure (U) as a result of loading, despite only modest changes in DXA-derived bone mass measurements. Data from (7).

Often exercise intervention studies report either small gains or no detectable effect on BMC and/or aBMD, prompting the conclusion that exercise is ineffective for improving bone strength. Our study suggests that, in rats, even small changes in bone mass, which are marginally detectable by DXA, can significantly improve bone strength by favorably altering bone geometry. If confirmed in human populations, these data suggest aBMD may not be the best parameter for assessment of the effectiveness of exercise. Instead assessment should include some measure of bone shape and size. For instance, an estimate of the cross-sectional moment of inertia provides information about the rigidity and strength of a bone that cannot be obtained from aBMD.


Hert’s (3) pioneering studies conducted more than 30 yrs ago demonstrated that bone tissue responds to dynamic rather than static loading. This finding, which has been replicated numerous times in other animal models, provides important information about how bone cells detect mechanical loading. Because static loads (even those that produce fairly large stresses or strains) do not initiate osteogenesis, stress or strain in the tissue cannot by itself be the primary stimulus for cellular response. Instead, there must be something special about dynamic loading.

Dynamic loading creates hydrostatic pressure gradients within bone’s fluid-filled lacunar-canalicular network. As those pressure gradients are equilibrated via movement of extracellular fluid from regions of high pressure to regions of low pressure, shear stresses are generated on the plasma membranes of resident osteocytes, bone lining cells, and osteoblasts. Bone cells are highly sensitive to fluid shear stresses and respond by initiating a cascade of cellular events, including elevation of intracellular calcium, paracrine/autocrine secretion, expression of growth factors, and, ultimately, bone matrix protein production. High-impact exercises that produce large rates of deformation of the bone matrix best drive fluid through the lacunar-canalicular network system. In addition, loading applied at higher frequency (cycles per second) more effectively stimulates osteogenesis. The required mechanical load necessary to initiate new bone formation decreases as the loading frequency increases. In a recent study in our laboratory in which controlled loading was applied to the rat ulna at 1, 5, and 10 Hz (cycles per second), the peak bone strain necessary to initiate osteogenesis in the cortex dropped from 1820 μstrain at 1 Hz to 650 μstrain at 10 Hz (Fig. 2) (4). Thus, increasing loading frequency is one step toward more effective application of mechanical forces to promote osteogenesis.

Figure 2
Figure 2:
Periosteal bone formation rate (rBFR/BS) measured in rat ulnae subjected to loading at different peak magnitudes and frequencies. The magnitude of load was converted to peak compressive strain at the medial surface of the ulna. The threshold strain necessary to initiate new bone formation was reduced from 1820 microstrain at 1-Hz loading frequency to 650 microstrain at 10 Hz. (Reprinted from Hsieh, Y-F., and C. H. Turner. Effects of loading frequency on mechanically induced bone formation.J. Bone Miner. Res. 16:918–924, 2001. Copyright © 2001 American Society for Bone and Mineral Research. Used with permission.)

This approach was recently tested in sheep (10). Small loads were applied to the sheep hindlimbs at 30 Hz for 20 min·day−1 for 1 yr. Trabecular bone volume was increased by more than 30% in the femur (one of the bones subjected to loading), but BMD was not increased in the loaded bones. Furthermore, there was no increase in bone mass at any cortical bone site, suggesting that the anabolic effect was limited to trabecular bone tissue. Hence it is unclear how effective high-frequency loading will be with regard to improving bone strength. Cortical bone contributes most to the structural integrity of the long bones and the hip, so it remains important to devise a high-frequency loading regimen that improves cortical bone mass and bone strength.


Mechanoreceptor cells that sense touch and sound become desensitized to the prolonged stimulus, and consequently, the signal is not processed by the central nervous system (and no sensation is experienced). Bone cells also exhibit a desensitization phenomenon in the presence of extended mechanical-loading sessions. Rubin and Lanyon (9) showed the osteogenic response to mechanical loading was not increased when the loading regimen was lengthened from 36 to 1800 consecutive cycles·d−1, suggesting that bone tissue rapidly becomes desensitized to prolonged exercise (Fig. 3). Others have replicated these findings. Notably Umemura et al. (14) showed that rats trained to jump multiple times·d−1 increased the mass of their femora and tibiae, but the anabolic response saturated after about 40 loading cycles; animals trained to jump 100 times·d−1 did not improve their bone mass significantly over those trained to jump 40 times·d−1. The data from Umemura et al. almost perfectly fit a logarithmic relationship, indicating that: where N is the number of jumps or loads on the bone during a training session. Equation 1 also demonstrates that bone tissue sensitivity to mechanical loading is proportional to 1/(N + 1). Thus, bone loses more than 95% of its mechanosensitivity after only 20 loading cycles. Presumably, bone cell mechanosensitivity will return after a period of no loading.

Figure 3
Figure 3:
Bone mass in the tibia of rats (closed circles; Umemura) or ulna of turkeys (open triangles; Rubin and Lanyon) increases after applied mechanical loading. However, the anabolic effect of loading saturates as the number of loading cycles increases. There is limited benefit of additional loading cycles above approximately 40 cycles·d−1. (Reprinted from Burr, D. B., A. G. Robling, and C. H. Turner. Effects of biomechanical stress on bones in animals. Bone. 30:781–786, 2002. Copyright © 2002 Elsevier Science. Used with premission.)

This concept was tested recently by Robling et al. (8) in an experiment in which rats were subjected to mediolateral bending of the tibia. All six groups of rats in the experiment received 360 cycles·d−1 of the same loading stimulus, delivered as 90 continuous cycles, 4 times·d−1. Some of the rats were allotted 8 h between each of the 4 daily bouts, others were given 4, 2, 1, 0.5, or 0 h (equivalent to 360 cycles continuously) of rest between each of the 4 daily bouts. Load-induced bone formation was improved by the rest periods, and as the rest (no loading) periods were lengthened, bone formation was enhanced further. The data from Robling et al. fit the following relationship:where t is the time between bouts and τ is a time constant approximately equal to 6 h (Fig. 4). With a rest period of 4 h between loading bouts, loading-induced bone formation was almost doubled. After 24 h of rest, 98% of bone mechanosensitivity returns. Consequently, the osteogenic response to exercise can be enhanced by regimens that incorporate periods of rest between short vigorous skeletal-loading sessions.

Figure 4
Figure 4:
Bone tissue desensitizes to mechanical loading rapidly. As loading cycles (N) increase, the mechanosensitivity decreases as 1/(N + 1). After loading is stopped, the mechanosensitivity recovers following the relationship 1−e−t/τ, where τ is approximately 6 h. Consequently, exercise is most effective if delivered in short bouts separated by several hours.


Equations 1 and 2 form the foundation for assessing the OI of exercise. The OI for a single session of exercise is intensity × ln (N + 1). The intensity of skeletal exercise is best defined by the loads applied to the bone. Laboratory data suggest that exercise intensity should be calculated as peak magnitude of load (or stress) multiplied by the loading frequency (13). Translating the laboratory data in terms of human exercise requires that we measure the magnitude of force on the bone and the duration of the applied force. For many exercises, the magnitude of force on the skeleton is proportional to the ground reaction force measured using a floor-mounted force plate. For a given exercise regimen, intensity of loading may be estimated as peak ground reaction force divided by the duration of the ground reaction force. Because the duration of the ground reaction force is seldom reported in exercise studies, we will use only the peak reaction force as an estimate of intensity in our calculations. For instance, the weekly OI generated by 20 min of walking 5 d·wk−1 is 36.8. This is calculated under the assumption that a 20-min walk generates approximately 800 loading cycles to each leg and that the peak load is 1.1 times body weight, or EQUATION

The OI for multiple bouts of loading depends upon the recovery time allowed between sessions. Therefore, three parameters are required to assess OI: intensity, N, and time between sessions. For mild-impact exercise, such as jumping, 5 times·wk−1, the OI varies from 50 when jumping 150 times·wk−1 to 70 for 600 jumps·wk−1. The osteogenic potential of exercise can be increased further when the daily exercise is divided into two shorter sessions separated by 8 h. For example, consider 120 jumps·d−1 done in one session or broken into either two sessions of 60 jumps separated by 8 h or three sessions of 40 jumps separated by 4 h:MATHMATHMATH

Consequently, breaking 120 jumps into two sessions improves the OI by almost 50%, but dividing the day further into three sessions 4 h apart does not further improve the OI. Calculation of OI for weekly exercise regimens demonstrates that the OI is best improved by adding more exercise sessions·wk−1, rather than lengthening the duration of individual sessions (Fig. 5). For instance, 300 jumps·d−1 done 2 times·wk−1 generates an OI of 33 whereas 120 jumps·d−1 done 5 times·wk−1 produces an OI of 70. Thus the osteogenic effectiveness of 600 jumps·wk−1 is more than doubled if the exercise is delivered in 5 daily sessions, rather than 2 times·wk−1. Short intense exercise bouts build bone most effectively; hence running short sprints should build more bone than a long jog. If one wishes to reduce exercise time, it is far better to shorten each exercise session than to reduce the number of sessions, as is illustrated by recent studies. Jump training improved BMC in the hip and spine when performed 3 times·wk−1 (1). However, when the number of sessions was reduced to 2 times·wk−1, jumping did not significantly affect BMC (2).

Figure 5
Figure 5:
Osteogenic index (OI) calculated for mild-impact loading (peak force is three times body weight). For the same number of loading cycles·wk−1, the OI is increased over threefold if the exercise is administered 5 d·d−1 rather than 1 ·d−1. The OI is increased by as much as 50% if the daily exercise is divided into 2 shorter sessions separated by 8 h (5 d·wk−1 × 2). Separating the exercise into 3 sessions 4 h apart (5 d·wk−1 × 3) does not improve OI beyond that achieved with 2 exercise sessions·d−1 (5 d·wk−1 × 2). The dashed line indicates the OI for walking for 20 min·d−1 5 d·wk−1.


“Old age is like everything else. To make a success of it, you’ve got to start young.”—Fred Astaire

Although exercise has clear benefits for the skeleton, engaging in exercise during skeletal growth is unequivocally more osteogenic than exercise during skeletal maturity (Fig. 6). The biological mechanisms for this phenomenon are not yet fully understood but are probably related to the fact that during growth, the bone surfaces are covered with greater proportion of active osteoblasts compared with the same surfaces after skeletal maturity. Consequently, the recruitment step of osteoblastic bone formation (migration of precursors to the bone surfaces) often is not necessary when the growing skeleton is mechanically stimulated; the mechanical stimuli can have direct effects on the osteoblastic cell populations already in situ. An additional and perhaps more compelling reason to engage in exercise during growth is that periosteal expansion occurs predominantly during growth, and consequently, the childhood and adolescent years provide a window of opportunity to enhance significantly periosteal growth with exercise. Periosteal growth determines the periosteal breadth of a bone, which is important in skeletal health for two main reasons. First, addition of bone to the periosteal surface improves the bending and torsional strength of the bone most effectively (see Fig. 1). Second, resorption of bone from the periosteal surface is extremely rare in the adult; usually it is the trabecular, endocortical, and Haversian bone surfaces that undergo remodeling. These observations indicate that the periosteal breadth of a bone will remain intact until senescence. A study conducted in growing rats trained to jump to an elevated platform 40 times·d−1 for 4 wk reported a significant increase in periosteal perimeter, which was maintained after a further 4 wk of normal activity (11). The same phenomenon has been reported in elite female tennis and squash players who have decreased their training. Collectively, these studies suggest that exercise during growth can significantly and irreversibly enhance periosteal bone growth and bone strength. Consequently, vigorous exercise during growth and young adulthood may well reduce fracture risk in later decades (6).

Figure 6
Figure 6:
Mechanical loading presents a much greater stimulus to the growing skeleton than it does to the mature skeleton. A study of humeral BMC in competitive female tennis and squash players showed that those who started playing at an early age (several years before menarche) had more than two times as much differential (playing arm vs nonplaying arm) in mineral accrual than those who started playing during their adult years. Data from (5).

In elderly adults with low bone mass, exercise constitutes only a moderately effective bone-building therapy. A meta-analysis of controlled exercise trials likened the skeletal benefits of exercise to what might result from calcium supplementation, that is, a modest reduction of bone resorption resulting in only minor reduction in BMD loss per year (15). However, it is important to recognize that exercise can effectively reduce fracture risk even without dramatic effects on bone mass. In addition, the key to reducing many osteoporotic fractures is protecting the skeleton from trauma by reducing the frequency of falls. Proper exercise can reduce falls by improving balance and postural stability. Exercises that increase postural stability and/or flexibility (e.g., Tai Chi and Yoga) appear to improve balance and one’s ability to recovery from a stumble, and thus lower the risk of dangerous falls.


1. Fuchs, R. K., J. J. Bauer, and C. M. Snow. Jumping improves hip and lumbar spine bone mass in prepubescent children: A randomized controlled trial. J. Bone Miner. Res. 16: 148–156, 2001.
2. Fuchs, R. K., D. P. Williams, and C. M. Snow. Response of growing bones to a jumping protocol of reduced repetitions: A randomized controlled trial. J. Bone Miner. Res. 16 (Suppl): 203, 2001.
3. Hert, J., M. Lisková, and J. Landa. Reaction of bone to mechanical stimuli. 1. Continuous and intermittent loading of tibia in rabbit. Folia Morphol. (Praha). 19: 290–300, 1971.
4. Hsieh, Y-F., and C. H. Turner. Effects of loading frequency on mechanically induced bone formation. J. Bone Miner. Res. 16: 918–924, 2001.
5. Kannus, P., H. Haapasalo, M. Sankelo, H. Sievanen, M. Pasanen, A. Heinonen, P. Oja, and I. Vuori. Effect of starting age of physical activity on bone mass in the dominant arm of tennis and squash players. Ann. Intern. Med. 123: 27–31, 1995.
6. Karlsson, M. K., H. Ahlborg, K. J. Obrant, F. Nyquist, H. Lindberg, and C. Karlsson. Exercise during growth and young adulthood is associated with reduced fracture risk in old ages. J. Bone Miner. Res. 17 (Suppl 1): S297, 2002.
7. Robling, A. G., F. M. Hinant, D. B. Burr, and C. H. Turner. Improved bone structure and strength after long-term mechanical loading is greatest if loading is separated into short bouts. J. Bone Miner. Res. 17: 1545–1554, 2002.
8. Robling, A. G., D. B. Burr, and C. H. Turner. Recovery periods restore mechanosensitivity to dynamically loaded bone. J. Exp. Biol. 204: 3389–3399, 2001.
9. Rubin, C. T., and L. E. Lanyon. Regulation of bone formation by applied dynamic loads. J. Bone Joint Surg. 66-A: 397–402, 1984.
10. Rubin, C., A. S. Turner, S. Bain, C. Mallinckrodt, and K. McLeod. Anabolism. Low mechanical signals strengthen long bones. Nature. 412: 603–604, 2001.
11. Singh, R., Y. Umemura, A. Honda, and S. Nagasawa. Maintenance of bone mass and mechanical properties after short-term cessation of high impact exercise in rats. Int. J. Sports Med. 23: 77–81, 2002.
12. Snow, C. M., D. P. Williams, J. LaRiviere, R. K. Fuchs, and T. L. Robinson. Bone gains and losses follow seasonal training and detraining in gymnasts. Calcif. Tissue Int. 69: 7–12, 2001.
13. Turner, C. H. Three rules for bone adaptation to mechanical stimuli. Bone. 23: 399–407, 1998.
14. Umemura, Y., T. Ishiko, T. Yamauchi, M. Kurono, and S. Mashiko. Five jumps per day increase bone mass and breaking force in rats. J. Bone Miner. Res. 12: 1480–1485, 1997.
15. Wallace, B. A., and R. G. Cumming. Systematic review of randomized trials of the effect of exercise on bone mass in pre- and postmenopausal women. Calcif. Tissue. Int. 67: 10–18, 2000.

bone mass; bone strength; exercise; biomechanics; osteoporosis; bone fracture

©2003 The American College of Sports Medicine