Field-based assessments provide a cost–effective and accessible alternative to dual-energy X-ray absorptiometry (DXA) for practitioners determining body composition in athletic populations. It remains unclear how the range of physical impairments classifiable in wheelchair sports may affect the utility of field-based body composition techniques. The present study assessed body composition using DXA in 14 wheelchair games players who were either wheelchair dependent (non-walkers; n = 7) or relied on a wheelchair for sports participation only (walkers; n = 7). Anthropometric measurements were used to predict body fat percentage with existing regression equations established for able-bodied persons by Sloan and Weir, Durnin and Womersley, Lean et al, Gallagher et al, and Pongchaiyakul et al. In addition, linear regression analysis was performed to calculate the association between body fat percentage and BMI, waist circumference, sum of 6 skinfold thickness and sum of 8 skinfold thickness. Results showed that non-walkers had significantly lower total lean tissue mass (46.2 ± 6.6 kg vs. 59.4 ± 8.2 kg, P = 0.006) and total body mass (65.8 ± 4.2 kg vs. 79.4 ± 14.9 kg; P = 0.05) than walkers. Body fat percentage calculated from most existing regression equations was significantly lower than that from DXA, by 2 to 9% in walkers and 8 to 14% in non-walkers. Of the anthropometric measurements, the sum of 8 skinfold thickness had the lowest standard error of estimation in predicting body fat content. In conclusion, existing anthropometric equations developed in able-bodied populations substantially underestimated body fat content in wheelchair athletes, particularly non-walkers. Impairment specific equations may be needed in wheelchair athletes.