This study followed up a cohort of patients with microalbuminuria identified on a spot morning urine sample 8 years earlier and aimed to determine if a spot morning urinary albumin concentration was able to identify patients with non-insulin treated diabetes at increased risk of mortality and progression to nephropathy. In 1984, 47 of 216 patients chosen by random selection from our teaching hospital-based diabetes clinic were identified as having microalbuminuria (urinary albumin concentration 35-300 micrograms ml-1). Subjects were compared with an age-matched control group from the 1984 cohort who did not have microalbuminuria. Eight years later, 22 of 47 (46.8%) patients with microalbuminuria had died compared to 10 of 47 (21.3%) patients without albuminuria (p < 0.05). The majority of deaths were from cardiovascular disease (53.1%). Logistic regression showed microalbuminuria to be an independent predictor of mortality, not influenced by age, duration of diabetes, blood pressure, glycosylated haemoglobin or creatinine at the initial examination. Eight years later, in the group with initial microalbuminuria, eight still had microalbuminuria and five patients had developed nephropathy. In the group without albuminuria in 1984, only one patient had progressed to microalbuminuria and no patients to nephropathy. In conclusion, a spot urinary albumin concentration is of value in identifying patients with an increased risk of mortality or progression to nephropathy, and is simple to obtain at a clinic.