Calculation of creatinine generation rate (CGR) has been reintroduced as a method to estimate lean body mass (LBM) in dialysis patients. It has also been suggested that it be used to identify noncompliance with dialysis prescription. In order to evaluate this method, LBM calculated from CGR (LBMCGR) was compared to 48 simultaneous estimations of LBM from measurements of total body potassium (TBK) in 35 patients on peritoneal dialysis (PD). TBK (mmol) was measured in a whole body counter and LBM (kg) was calculated as TBK/68.1 (Forbes). CGR was calculated with and without inclusion of "metabolic degradation" of creatinine. LBMCGR was further estimated using two different equations based on results in healthy subjects, one from a group on an ad libitum diet, the other from a group on a meat-free diet. The intercept of the two equations differs by 13 kg. CGR systematically underestimated LBM when compared to TBK, but to a lesser degree when using the equation based on a meat-free diet. Repeated determinations of CGR in 11 stable patients revealed an unacceptably high coefficient of variation (CV%) for LBM, 14.2%, while body wt CV was 1.4%. We conclude that CGR is not a valid method to monitor LBM in PD patients. CGR is highly variable and in part dependent on meat intake, as is the relation between CGR and LBM. For the same reasons, it seems unwise to draw conclusions of noncompliance of PD-patients from determinations of CGR.