Timed urine collections from diabetic children and adolescents were assessed for urinary albumin excretion rate (microgram min-1) before and after freezing at -20 degrees C. Freezing had the effect of changing the estimation such that frozen values could differ from that of the fresh, by as much as one half to twice as much. The variation depended on the concentration defining the initial albumin excretion rate but was not influenced by the length of storage when frozen. We conclude that researchers should be aware that freezing and storing of urine samples prior to albumin concentration assessments can affect the absolute values obtained. It would appear more appropriate to analyse samples prior to freezing to be certain of obtaining true prevalence estimates of microalbuminuria.