The urea kinetic model (UK) and the direct dialysis quantification method based on dialysate collection (DDQ) were used to determine the urea distribution volume (V) identified with the total body water and the urea generation rate (G) for different dialysis times, both in vivo during short hemodialysis (N = 20) and in vitro using an experimental single-pool urea system (N = 10). Both UK and DDQ allowed a satisfactory in vitro estimation of V and G for all dialysis times. On the other hand in vivo V and G estimations by both methods showed an increase of more than 50% between the determinations performed after 30 minutes of dialysis and at the end of dialysis. Our theoretical analysis shows that the in vivo changes of V are compatible with those expected for a two-compartment system in which one compartment is cleared faster than the other. Furthermore, given that urea is allowed to equilibrate in the body at the end of dialysis, DDQ permits an accurate estimate of V, G and PCR even for short hemodialysis, which UK does not.