The use of U.S. vital statistics for surveillance of drug-related mortality may be limited by the way in which certifiers complete death certificates and by the constraints of the International classification of diseases, Ninth Revision (ICD-9). ICD-9 is the system used by the National Center for Health Statistics (NCHS) to compile national, cause-specific mortality data from information reported on death certificates. To investigate the extent of variability in certification practices among medical examiners (MEs), we conducted a mailout survey in which we asked a national sample of 49 MEs to review summaries of 28 death scenarios and, for each death, assign the cause and manner of death. Cocaine use was the unequivocal cause of death for 17 of the 28 deaths. We then asked a nosologist at NCHS to code the verbatim survey responses in accordance with the rules and rubrics of the ICD-9 system. Of the 20 MEs who responded, 14 provided complete cause and manner determinations. For the cocaine-caused deaths, the 14 respondents provided 238 cause-of-death statements; 220 (92.4%) explicitly mentioned cocaine. However, only 45 of the 238 responses (18.9%) led to a cocaine-specific ICD-9 code for the underlying cause of death. Our findings illustrate how death certification practices, coupled with the ambiguities of the ICD-9 system, may lead to substantial loss of detail about cocaine-caused deaths and misclassification of these deaths in official compilations of mortality statistics.