Two normal control populations, separated by 8,000 miles and 24 degrees of latitude, had similar six-month mean values for overnight urinary melatonin concentrations. These values were significantly higher than six-month values for depressed subjects and abstinent alcoholic subjects, while the means for the two clinical populations were similar. Age and urinary melatonin concentration in the control and clinical populations were inversely related, but the slopes of the linear regression equations were ten times steeper for the control populations than for the clinical populations. Differences in age and sex distributions accounted for some of the differences in values between controls and the clinical populations, although controls still differed from the clinical populations, even after sex and age were factored out. The disparate slopes for age and melatonin concentrations may contribute to some of the conflicting findings of studies comparing populations of different ages. The total melatonin content in the samples from alcoholic subjects, but not the depressed subjects, was lower than that for controls. The difference in the urinary melatonin concentration between the controls and the two patient groups was not accounted for by difference in duration of urine collection period, hours of sleep or body weight.