Fifty individuals with lower limb impairments including spinal cord injury, polio and amputations underwent aerobic and anaerobic arm-cranking tests in a standardized laboratory setting. Based on linear regression models applied with age as dependent variable aerobic performance variable including HRmax (R=0.395, P=0.004), and POaer (R=0.31, P=0.021) were subjected to ANCOVA adjusting for age in order to determine the signi®cance of participation intensity (competitive vs leisure) and type of physical impairment. Anaerobic performance variables were not in¯uenced by age and thereby subjected to 1-Way ANOVA with the same independent variables. Participation intensity and type of impairment signi®cantly discriminated (P50.001) between athletes in all power variables. Linear regression models have shown moderate but signi®cant (P50.001) relationship with functional ability (bases on International Wheelchair Basketball Federation classi®cation system). In anaerobic mean power (MP) classi®cation accounted for 42% of the variance, while in anaerobic peak power (PP) and aerobic Power (POaer) for 38% and 30% respectively. By means of a post hoc Tukey analysis signi®cant dierences were observed between athletes with a high level paraplegia (class 1) and those with one leg aected by polio or amputations (classes 4, 4.5). Athletes with low level paraplegia and two legs aected by polio (classes 2 ± 3.5) had values in-between. Based on the descriptive evaluation, a three group scheme was conceptualized and resubjected to ANOVA. Signi®cant intergroup dierences were thus obtained only for PP. Descriptive PP data for each group were transformed into a ®ve category table in order to provide reference values for ®tness estimation in males with lower limb impairments of various etiologies.