The aim of this component of the German Study on Sudden Infant Death was to determine (1) nicotine concentrations in hair (NCH), as a marker of long standing exposure to tobacco, (2) cotinine concentrations in pericardial fluid (CCP) and (3) cotinine concentrations in liquor cerebrospinalis (CCL), the latter measures being markers of recent exposure to tobacco in the last few hours of life. The results obtained were compared with data on parental smoking revealed from interviews. In 100 cases of sudden infant death syndrome, material was taken at autopsy to determine NCH. In 41 cases, NCH and CCP, and in 70 cases, NCH and CCL were determined. Infants of mothers who stated having smoked during pregnancy had higher NCH than infants of non-smoking mothers (p = 0.008). Furthermore, there was a weak but statistically significant relationship between NCH’s and the daily cigarette consumption of the mother during pregnancy (n = 64, r = 0.24, p = 0.05). In 43% of infants, nicotine could be detected in their hair, although the mothers had said at the interview that they did not smoke during pregnancy. On the other hand, in 33% of infants whose mother stated they had smoked during pregnancy nicotine was not detectable in the infant’s hair. CCP’s were strongly correlated with CCL’s (r = 0.62, p = 0.0027). For this reason, both parameters were treated as equivalent for the detection of tobacco smoke exposure in the last hours before death. The influence of breast-feeding was evaluated by comparison of the nicotine concentrations in breast fed and non-breast-fed infants from smokers and non-smokers. Fivefold higher nicotine concentrations were determined in non-breast-fed infants of parents who smoked as compared to all other groups. It can be concluded that nicotine intake by passive smoking is much more important than by breast-feeding. We conclude that both interview data and biochemical measures should be sought to understand the true exposure to tobacco smoke.