It has been claimed that computed urea kinetic (UK) modelling in hemodialysed patients, for the estimation of protein intake, leads to an overestimation of protein catabolic rate (PCR). In the present study, three different methods of kinetic modelling for the determination of PCR and Kt/V are compared in 24 patients. The first method was the direct quantification method (DDQ) based on the collection of all urea eliminated from the body. The first computed method (ICMI) was the urea kinetic modelling method as described by Sargent. Dialyzer clearances were measured directly and not estimated by theoretical extrapolation. The second computed method (ICMII) is based on the indirect calculation of urea distribution volume (Vu), according to Watson, and of dialyzer clearances from this Vu and from pre- and post-dialysis urea concentrations. All three methods resulted in PCR's that were not significantly different (DDQ: 1.03 +/- 0.19; ICMI: 1.04 +/- 0.22; ICMII: 1.08 +/- 0.25 mg/Kg BW.24 hrs; p greater than 0.05). When the results were correlated, the following results were obtained: ICMI vs ICMII: r = 0.89, p less than 0.001; ICMI vs DDQ: r = 0.68, p less than 0.01; DDQ vs ICMII: r = 0.78, p less than 0.001. Intermutual comparison of Kt/V values resulted in virtually identical results, especially when comparing ICMI and ICMII, where the regression line equalled the identity line. In conclusion, all methods seem equally reliable in determining mean PCR and Kt/V. Our data, obtained with directly measured dialyzer urea clearances, do not confirm the earlier held opinion that computed modelling results in an overestimation of PCR.(ABSTRACT TRUNCATED AT 250 WORDS)