Serum/plasma urea (S/P-urea) and creatinine (S/P-creatinine) concentrations are routinely assayed as indirect markers of glomerular filtration rate and have been reported to be highly correlated. The aim of this study was to evaluate the relationship between the two analytes in a large number of unselected samples submitted to a commercial laboratory. In 4799 pairs of results, the correlation was high (r = 0.795) and was not affected by sex or age. The relationship between the two analytes was best represented by a second-order polynomial equation. However, the dispersion of results was large and there was a high percentage of cases (27.5%) where S/P-urea was increased and S/P-creatinine normal (< or =120 micromol/l for this laboratory), while there was a low percentage of cases where S/P-creatinine (1.6%) was increased and S/P-urea normal (< or =8 mmol/l for this laboratory). The discrepancy between increases in S/P-urea and S/P-creatinine might not only reflect a high incidence of non-renal factors of variation for S/P-urea but also an effect of the size or muscle mass of the dogs on the limit of decision for S/P-creatinine. In dogs with normal S/P-urea, there was a significant effect of the size on the 0.975 quantile of S/P-creatinine, ranging from 106 micromol/l in very small dogs to 133 micromol/l in large and very large dogs. This study shows that isolated increases in S/P-urea could be misleading for the diagnosis of renal diseases and that the reference intervals of S/P-creatinine should be re-evaluated according to breed or muscle mass of dogs.