Aim. (a) To determine the normalized cellular clearance (K(c)n) of urea, creatinine and phosphate in patients undergoing maintenance hemodialysis; (b) To identify the factors, particularly circulatory, which determine K(c)n; (c) To evaluate whether intra-dialytic blood sampling can predict the size of the post-dialytic solute concentration rebound. Methods. Kinetic modelling of urea, creatinine and phosphate, using a two-pool variable volume computer simulation, was performed on two occasions on 34 patients undergoing maintenance dialysis. The cellular clearance was determined (a) from the size of the rebound 50 min after the end of dialysis; (b) from a mid-dialytic blood sample. Conventional two-dimensional M-mode echocardiography and Doppler peripheral blood pressure measurement were performed. Results. The model produced accurate measurements of rebound K(c) for urea in 93% of measurements, creatinine in 49% and phosphate in 13%. The corresponding figures for mid-dialysis K(c)n were 76%, 39% and 0%. The rebound K(c)n was, for urea, 8.31 ± 4.31 ml/kg/min, and for creatinine 4.07 ± 2.98. The mid- dialysis K(c)n was, for urea, 8.57 ± 4.25 ml/kg/min, and for creatinine 5.06 ± 3.36. High post-dialytic rebounds (and low K(c)n values) were associated with erythropoietin use (p < 0.05) and occurrence of end-dialytic hypotension (p < 0.02). Patients treated with calcium antagonists had a significantly (p < 0.001) higher K(c)n. There was no correlation between mid-dialysis and rebound K(c)n. Circulatory indices had no influence on K(c)n. Conclusions. The two-pool cellular clearance model is compatible with urea kinetics, but not creatinine or phosphate. It is therefore unlikely that it is the correct model for small molecule kinetics. The post-dialytic solute rebound may be partly an iatrogenic phenomenon and can be reduced by preventing post- dialytic hypotension and by calcium antagonist treatment, both of which improve regional blood flow. The size of the rebound cannot be predicted by intra-dialytic blood sampling.