BACKGROUND. Microalbuminuria may predict proteinuria and increased mortality in non-insulin dependent diabetic patients. Early detection of microalbuminuria may therefore be essential. AIM. The primary objective of this study was to describe the association between the presence of albuminuria in diabetic patients as detected by general practitioners using conventional reagent strip dipstick tests for albumin, and the urinary albumin concentration as measured in a hospital laboratory. METHOD. A total of 675 newly diagnosed diabetic patients aged 40 years or over were included in the Danish study, diabetes care in general practice. Data for urinary albumin concentration from a morning urine sample and the results of three consecutive dipstick tests for albumin were collected for 417 patients. RESULTS. When defining elevated urinary albumin concentration as 200 mg l-1 or more (proteinuria) the finding of at least one positive test out of the three dipstick tests for albumin had a diagnostic sensitivity of 73% and a specificity of 89%. When the microalbuminuric range (15.0 to 199.9 mg l-1) was added to the definition of renal involvement, the sensitivity of the dipstick test became as low as 28% with a specificity of 96%. CONCLUSION. It is essential for general practitioners to be able to identify proteinuric patients. To achieve this by means of the conventional dipstick test, general practice procedures need to be improved. As it is becoming increasingly well-documented that microalbuminuric non-insulin dependent diabetic patients may benefit from pharmacological treatment of even slight arterial hypertension and heart failure, it seems reasonable to suggest that the use of dipsticks for albumin in general practice be replaced by laboratory quantitative determination of urinary albumin concentration in a morning urine sample.