A method for the estimation of time of death (TOD), was evaluated by measuring the melatonin (MT) content of pineal bodies (PBs), sera and urine samples from 85 cadavers. A total of 44 cadavers were investigated in Sapporo (geographical coordinates N 43 degrees 4', E 141 degrees 21') and 41 in Tokyo (N 35 degrees 39', E 139 degrees 44'). MT contents were measured by radioimmunoassay (RIA) in 75 PBs, 27 sera and 14 urine samples. Exponential differences of pineal MT content were recognized between peaks in nighttime and nadirs in daytime, ranging from 0.099 to 63.2 ng/PB. Circadian rhythms were also observed for the concentrations of MT in serum (11-205 pg/ml), and in urine (7.5-137.5 pg/ml). Consequently, criteria for the TOD estimation are proposed as follows. 1) Pineal MT contents--(1) 0-0.2 ng/PB: TOD 1100-1700 hours, (2) 0.2-0.3 ng/PB: TOD 0700-2000 hours, (3) 0.3-1 ng/PB: inconclusive, (4) 1-4 ng/PB: TOD 1600-1000 hours, (5) 4-8 ng/PB: TOD 2000-0800 hours, (6) over 8 ng/PB: TOD 2000-0500 hours, 2) Serum MT concentration--(1) 0-100 pg/ml: inconclusive, (2) over 100 pg/ml: TOD 2200-0100 hours, and 3) Urinary MT concentration--(1) 0-35 pg/ml: inconclusive, (2) over 35 pg/ml: TOD 1800-0600 hours. The range of the estimation can be limited by a combination of these 3 criteria. The present method can be combined with other methods for estimating the TOD to decrease the range.