In order to establish a method for the estimation of death time (DT) from measuring melatonin (MT) contents in pineal bodies (PBs) and biological fluids, 85 cadavers were investigated--44 dead in Sapporo (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 in 76 PBs, 27 sera and 14 urines. Exponential differences were recognized between peaks in nighttime and nadirs in daytime of pineal MT contents, i. e., ranging 0.099-63.158 ng per PB and 1.2-609.6 pg/mg. Circadian rhythms were also observed on the concentrations of MT in serum (11-205 pg/ml), and in urine (7.5-137.5 pg/ml). Consequently, criteria of the DT estimation were proposed as follows. 1) Pineal MT contents--(1) 0-0.2 ng/PB:DT 11:00-17:00, (2) 0.2-0.3 ng/PB:DT 7:00-20:00, (3) 0.3-1 ng/PB: incapable of DT estimation, (4) 1-4 ng/PB:DT 16:00-10:00, (5) 4-8 ng/PB:DT 20:00-8:00, (6) 8 ng/PB over: DT 20:00-5:00, 2) Serum MT concentration--(1) 0-100 pg/ml: incapable of DT estimation, (2) 100pg/ml over: DT 22:00-1:00, and 3) Urinary MT concentration--(1) 0-35pg/ml: incapable of DT estimation, (2) 35 g/ml over: DT 18:00-6:00. Furthermore, width of the estimation is able to be narrowed by the combination of these three criteria. The present method is very useful in narrowing the width of DT estimation which has been usually carried out by observing the progress of cadaveric phenomena on and in dead bodies.