Background: Changes in body composition in cancer patients during chemotherapy are associated with treatment related toxicities or mortalities. Thus, it is relevant to identify accessible, relatively inexpensive, portable and reliable tools for evaluation of body composition in cancer patients during the course of their treatments.

Objective: To examine relationships between single cross-sectional thighs magnetic resonance imaging (MRI), skeletal muscle mass (SM) as reference and multi-frequency bioelectrical impedance analysis (BIA) fat free mass (FFM) in patients with colorectal cancer undergoing chemotherapy.

Design: In an observational, prospective study we examine the relationships between single cross-sectional thighs MRI (T1-weighted (1.5 T) SM compared to FFM BIA (8-electrodes multi-frequency Tanita MC780MA)) and FFM skin-fold thickness (ST) (4-points (Harpenden, Skinfold Caliper)) and SM equation for non-obese persons from Lee et al. 2000 (L2000) (based on age, height, weight, sex and race). FFM and SM (kg) were calculated based on either area (MRI) or weight.

Results: 18 CRC patients (10 males and 8 females) with mean (SD) age 67 yr (6) were measured at baseline, and 13 were available for follow-up. BIA overestimated FFM kg for all 31 measurements with mean (SD) 18.0 kg (6.0) compared to the MRI. ST overestimated FFM kg with mean 12.4 kg (6.2) and L2000 underestimated SM kg in 18 measurements and overestimated in 13 with a total mean of −4.3 kg (6.8). Conclusions BIA and ST were the best alternatives to MRI as they showed constant and thereby correctable errors. The equation, L2000, carried the smallest average measurement error but it was non-constant.