The objective was to evaluate the applicability and reliability of an unbiased stereological computerised tomography (CT) method for estimating total human body (HB), skeletal muscle (SM) and adipose tissue (AT) volumes in groups of neuromuscular patients. In 10 neuromuscular patients HB, SM and AT volumes were estimated using systematic sampling on equidistant CT sections throughout the total body axis using a counting grid with systematically ordered intersection points. Each intersection point hitting HB, SM or AT represented a known volume dependent on intersection point distance and sum of section thickness and gap. Random and systematic intra- and interobserver errors for volume estimates were below 0.035. These errors were negligible to the coefficient of variation of the group mean, being 0.190 for HB, 0.323 for SM and 0.471 for AT. Even in the presence of intrafascicular and intramuscular fat in neuromuscular patients, unbiased and reliable quantification of HB, SM and AT is possible.