Background: The International Classification of Diseases (ICD) external codes (E codes) for drowning assist in determining the primary event leading to drowning, but do not alone allow the precise determination of the overall drowning rates. Aims: To analyze the sensitivity of the ICD E codes for drowning. To describe the pattern and trend of drowning deaths that are classified with E codes other than for drowning. Setting: Finland, 1969–2000. Methods: Mortality files of Statistics Finland were searched electronically using the injury codes (I codes) and E codes for drowning. Cross analysis of I and E coded drownings was performed to determine the rate and pattern of drowning cases classified with E codes other than for drowning. Time trends were calculated using the Poisson regression model. Results: Of 13 705 drowning deaths, 644 (4.7%) were not identified with the E codes for drowning. The great majority (n=547, 84.9%) of these cases were traffic accidents resulting in drowning. No significant time trends were found even after the introduction, in 1996, of the ICD 10th revision. Conclusions: In Finland, underestimation of overall drowning rates using the E code alone is less pronounced than in countries where similar studies have been performed. The relatively high rate of transport accidents resulting in drowning indicates a specific target for preventive countermeasures.