Forensic pathologists often state that homosexual homicides are more violent than those with heterosexual victims. Overkill or wounding far beyond that required to cause death is a frequently used descriptor of these deaths. We quantified number and extent of injuries between homosexual and heterosexual homicide victims to determine whether one group suffered more violence than the other. This case-control study involved 67 homosexual homicide victims and 195 age, race, and gender-matched heterosexual (control) homicide victims from the Broward County Medical Examiner's Office in Fort Lauderdale, Florida, between 1982 and 1992. The sexual preference of the victims was determined by police and medical examiner investigation. We compared the percentage of cases with multiple causes of death, mean number of injuries/case, and the mean number of different body sites with injuries/case between the two groups. The percentage of cases with multiple causes was 12% (n = 8) and 5% (n = 10) in the homosexual and heterosexual victims, respectively, with no significant statistical difference detected (Fisher exact test, two-tail, p = 0.088). The mean number of fatal sharp (5.4 vs 3.3, p < 0.05), blunt (9.0 vs 6.5, p < 0.05), and total (14.5 vs 6.5, p < 0.001) injuries/case was greater among the homosexual victims than among the heterosexual group. These differences were all statistically significant using the Mann-Whitney rank sum test. The mean number of different body sites with injuries/case was also greater (3.5 vs 2.4, p < 0.001) among the homosexual victims than among the heterosexual group. This difference was statistically significant using the Mann-Whitney rank sum test. Homosexual homicides are more violent than heterosexual homicides when one compares the mean of number injuries (fatal sharp, blunt, and total)/case and the extent of injuries on the body. The mean number of stab wounds wounds/case and percentage of cases with multiple causes of death were also greater among the homosexual group, but the differences were not statistically significant.