Suicide statistics are generally recognised to be unreliable. This study of the reported rates of suicide in West Malaysia between 1966-1990 shows that the mean crude suicide rate between 1966-1974 was 6.1 per 100,000, but had dropped drastically between 1975-1990 to a mean of 1.6 per 100,000. Three lines of evidence are presented to show that this reduction in the suicide rate is due to a systematic misclassification of medically certified suicides as deaths due to undetermined violent deaths (which refers to violent deaths not known to be accidentally or deliberately inflicted). Firstly, the large drop in reported suicide rates corresponds closely to an increase in the rate of deaths due to undetermined violent deaths. There is a highly positive negative correlation between the two rates (coefficient of correlation, r = -0.9). Secondly, the misclassification appears to be mainly a problem with the medically certified deaths which follow the ICD classification. The mean ratio of uncertified to certified suicides before 1975 was 0.8, but from 1975 onwards the mean was 3.1. This is in contrast to the corresponding ratio for deaths due to all accidents which has remained fairly constant throughout these years. Thirdly, the race and sex differences for the rates of undetermined violent deaths are identical to those of suicide. Taking the misclassification into account the corrected suicide rate for West Malaysia is estimated to be between 8-13 per 100,000 since 1982.