SUMMARY. Salivary creatinine concentrations are 10-15% of serum creatmme concentrations in healthy populations but have not previously been measured in patients with renal disease. The Cobas Mira automated Jaffe method was adapted to measure salivary creatinine concentration. The method was linear to 1200/lmol/L, had a mean recovery of 78% and a detection limit of 6/lmol/L. Intra-assay variability was 13·5, 5·5 and 1·4% at 10'6, 17·8 and 128·4/lmo1/L, respectively. Interassay variability was 35·0 and 4·8% at 6·2 and 130·4/lmol/L, respectively. The median salivary creatinine concentrations was 8·5 /lmo1/L (range 6-18/lmol/L) in healthy subjects (n=23), and 84/lmo1/L (range 18-591 /lmol/L) in patients with renal disease (n = 25). Salivary and serum creatinine concentrations were not related in healthy subjects, however, a significant relationship was found in the patients (r = 0'784, P< 0·001). Using salivary creatinine concentration of 16·8 /lmol/L as a cut off value, all patients would have been detected with one false positive result (sensitivity 100%, specificity 95·7%, efficiency 97'7%). Thus, salivary creatinine estimations may be used to identify subjects with serum creatinine concentrations above 120/lmol/L.