To assess the possible relationship between maternal drug use during pregnancy and subsequent sudden infant death syndrome (SIDS), we identified 1760 cases of SIDS from a population of more than 1.2 million infants (1.45/1000) born in New York City between 1979 and 1989. The SIDS rate in drug-exposed infants was 5.83 per 1000 infants, compared with 1.39 per 1000 infants who were not drug exposed. With control for known associated high-risk variables, the risk ratio for SIDS in each individual drug group (methadone, 3.6; heroin, 2.3; methadone and heroin, 3.2; cocaine, 1.6; cocaine and methadone or heroin, 1.1) was higher than in the non-drug-exposed group. Higher rates of SIDS were found in infants exposed to opiates alone than in cocaine-exposed infants, but increasing rates of SIDS in cocaine-exposed infants toward the end of the decade suggested that "crack" cocaine may be linked to these increasing rates. Declines in the overall rate of SIDS during the decade were observed for both the drug-exposed (11.28 to 4.09 per 1000) and the nonexposed groups (1.70 to 1.05 per 1000). Differences in rates of SIDS between major racial-ethnic groups in nonexposed infants were not apparent if the mothers used drugs during pregnancy. Seasonal variation and distribution of ages at time of SIDS death did not differ between the drug-exposed group and the nonexposed group, suggesting that drug-associated SIDS may provide clues as to the cause or causes of SIDS.