There are a variety of methodological problems with published studies of parental smoking and sudden infant death syndrome (SIDS), with over-control the most consistent and problematic. Nevertheless, even though this is likely to minimise the true magnitude of relationships, the results are consistent. There are five cohort studies with prospectively collected information on maternal smoking in pregnancy: all show strong and statistically significant relationships that were dose dependent-the more cigarettes the mother had smoked, the more at risk was the baby of SIDS. Similar results have been shown from the case-control studies in which information has been collected retrospectively from parents or birth certificates. There are data from several studies indicating that environmental tobacco smoke (ETS) is also important. Since it has not yet been possible to determine conclusively whether associations are with smoking (or ETS) during pregnancy or postnatally, it is concluded that both should be discouraged.