The frequency of urinary infection was determined using quantitative microbiology in 172 insulin-dependent diabetic patients repeatedly being tested for microalbuminuria over 18 months on at least six occasions. The point prevalence of urinary infection at first screening for microalbuminuria was 3%. Over the period of study, 20 of the patients (12%) showed evidence of urinary infection, defined as a pure growth of a recognized pathogen > 10(7)I-1. Infection was more common in women than men (20% vs 5%, p < 0.01) and was significantly associated with the presence of peripheral neuropathy (p < 0.05). Infection was not related to patient age, duration of diabetes, glycaemia, blood pressure, retinopathy or autonomic neuropathy. There were no significant within-patient differences in albumin excretion, glycaemic control or blood pressure in relation to the presence and absence of urinary infection. In only one patient (5%) did urinary infection significantly increase the urinary albumin excretion and this was associated with pyuria. We conclude that the presence of urinary infection does not apparently affect the measurement of urinary albumin excretion unless pyuria is present. Unless diabetic patients are symptomatic, examination of the urine for infection is probably unwarranted when testing for microalbuminuria.