Measurement of serum concentration of trypsin by RIA-Gnost Trypsin kit (Hoechst-Japan) was evaluated. The clinical usefulness of measuring serum trypsin level in diabetic patients was assessed. The measurement of trypsin using the radioimmunoassay (RIA) kit revealed good precision and reproducibility with intraassay error ranging from 3.6 to 5.5% in C.V. corresponding to mean trypsin concentration of 236.5-838.7 ng/ml and interassay error ranging from 8.1 to 11.1%. Tests for recovery and dilution were satisfactory for clinical use. Clinical materials included 35 normal subjects, 88 diabetics, 22 patients with liver diseases, 3 with acute pancreatitis, 7 with chronic pancreatitis and 3 with chronic renal failure. Serum trypsin concentration in normal controls was 157.6 +/- 59.9 ng/ml (m + 1 S.D.). Diabetic patients treated with diet therapy alone revealed serum trypsin level of 203.6 +/- 74.8 ng/ml (n = 50). In diabetics treated with sulfonil urea serum trypsin was 171.3 +/- 83.0 ng/ml (n = 25). In patients receiving insulin serum trypsin level was 90.5 +/- 49.0 ng/ml (n = 13). In patients with liver diseases, acute pancreatitis, chronic pancreatitis and chronic renal failure serum trypsin concentration were 236.9 +/- 88.0, 520.1 +/- 80.0, 113.0 +/- 75.6, and 2557 +/- 2771 respectively. Our results may indicate impaired pancreatic exocrine function in patients with severe diabetes mellitus. Increased serum trypsin level in diabetics treated with diet therapy may be due to stimulated excretion of trypsin resulted from restricted food intake. However, further study in larger number of patients is needed.