The existence of temporal clustering of suicides was analyzed by comparing the distribution of weekly numbers of suicides under age 20 in Japan (1982-1987) with a Poisson distribution. The linkage between newspaper reports about teenage suicides and these clusters of suicides was examined in an ecologic study. The following results were obtained: 1) The distribution of weekly numbers of suicides under age 20 was not consistent with a Poisson distribution. The observed numbers for weeks with clustering was higher than the expected numbers. Therefore it is suspected that teenage suicides was not an independent occurrence but clustered in specific weeks with clustering observed in April and between the 15th and 25th weeks in 1986. 2) The age and sex characteristics of those who committed suicide during the observed clusters every April did not differ from that of suicide victims for the rest of the year. The incidence ratio for suicide between 15th and 25th weeks in 1986 compared to the same period for other years was higher in girls than in boys, and higher in younger teenagers than older teenagers. Analysis by prefecture of suicides which occurred between the 15th and 25th weeks in 1986 showed high incidences for Hokkaido, Tohoku, Chugoku, and Shikoku blocks, and low incidences for the Kanto and Kinki blocks. 3) Twenty news articles reporting teenage suicides with headings much larger than four columns appeared in the Asahi Newspaper in 1982-87. Significant linkage between newspaper reports about teenage suicides and trend of weekly numbers of suicides in 1985 and 1986 was not observed. 4) Weekly numbers of suicides significantly increased following newspaper reports on January 23rd 1985 and on April 8th 1986, describing a suicide caused by battering, and a jumping suicide of a young singer who was popular among teenagers, respectively. However no increase in the number of suicides was observed following other newspaper reports.