The effect of water temperature on the outcome of nearly drowned children was studied retrospectively. All patients under 16 years of age, who required admission to the paediatric intensive care unit (PICU) or who died despite life support measures between January 1, 1985 and December 31, 1994 in Southern Finland, were included in the study. The authors created a Near Drowning Severity Index (NDSI) and an age-adjusted NDSIage as tools to evaluate the effect of submersion duration and water temperature on the outcome of nearly drowned children. The predictive performance of the NDSI and the NDSIage were compared with the duration of submersion to predict the outcome. Of the 48 patients 11 were female and 37 were male. Their ages ranged from 0.8 to 15 years and the median was 3.7 years. The submersion time ranged from 0.5 to 90 min and the median time was 6.3 min. The median temperature of water was 16 degrees C (range 0-37 degrees C). The NDSI had a sensitivity of 93.9% and specificity of 90.6% in predicting the outcome after submersion incidents. The inclusion of age (NDSIage) in the formula did not increase predictive performance of the NDSI. With a cut-off value of 10 min, the duration of submersion alone had a sensitivity of 96.6% and specificity of 89.5% in predicting the outcome. In conclusion the effect of a potentially beneficial rapid development of hypothermia by cold water on the outcome of nearly drowned children could not be proved.