In a prospective study, 15,110 childhood traumas were recorded by the Pediatric Surgery Service (CHUV, Lausanne) between January 1, 1990 and December 31, 1997. The exact clock hour when the injury occurred and other germane data were obtained. Time series thus obtained were analyzed by several statistical (ANOVA, cosinor, chi2, Table Curve, etc.) methods. High statistically significant circadian patterns were detected with a trough at night--almost no traumas/hour (t/h), and a peak in the afternoon (approximately 16:00h)--9.3 +/- 0.4 (SD) t/h. Such 24h variation was validated for the whole sample for the entire 8 yr study span as well as the data of each year. Neither gender- nor age-related differences in the 24h pattern were detected between children under 5 yr of age, who have not yet attended school and children from 5 to 16 yr of age, who attend school. Small but statistically significant differences in the 24h patterns were observed when categorized by the type of activity associated with the trauma and the place of trauma occurrence. The great stability of the 24h pattern in childhood trauma over the 8 yr study span suggests an endogenous origin in addition to the role presumably played by environmental factors. Periods of 12 and 8h were also detected in the time series. The afternoon peak time of childhood traumas differs from that of adults, which is located approximately 04:00h in rotating shift workers and automobile drivers and 06:00-08:00h in adult day-workers. The validation of a circadian pattern in childhood traumas with an afternoon peak should be taken into account in the design of children's preventative injury programs.