As in previous studies an improved single radial immunodiffusion technique was used for albumin, calculating the albumin/creatinine ratio in spot urine (mg/g/1.73 m2). This ratio was 4.7 +/- 0.0174 (geometric mean +/- SEM as logarithm) in 130 healthy controls, the highest value being 10.8. The 182 non-selected ambulatory diabetic patients presented three subgroups, each showing a non-gaussian frequency distribution: 47% with normal values (5.0 +/- 0.0224); 42% with ratio values from 11.0 to 88.8 (22.7 +/- 0.0269, significantly differing from the controls); 11% with clinical proteinuria (subsequently excluded from the study). Type I (9.6 +/- 0.0452; n = 76) and type II diabetic patients (10.7 +/- 0.0430; n = 86) significantly differed (p less than 0.001) from the controls but not from one another. Irrespective of the diabetes type, ratio values were significantly correlated with the duration of diabetes, age of patients, age of diagnosis (for instance 16.2 +/- 0.0974 in 15 patients aged greater than 65 years versus 9.1 +/- 0.0792 in 23 patients aged less than 20 years), glycemia level and chronic complications (especially retinopathy). Therefore, more than half the diabetic patients, non-selected, presented an increased albumin excretion as compared to the controls. On the other hand, microalbuminuria appears to be linked to age, duration of the disease and quality of the metabolic control rather than to the diabetes type.