In a hospital based case-control study, we investigated the role of environmental factors in the aetiology of Parkinson's disease. This paper describes our results on smoking habits. Methods. The smoking histories of 380 Parkinson's disease (PD) patients recruited from nine German clinics were compared to those of age- and sex-matched control subjects (379 neighbourhood controls and 376 controls from the same region). Detailed information on smoking behaviour was collected in structured personal interviews in order to calculate the number of pack-years smoked up to the time of diagnosis. Conditional logistic regression was used to calculate odds ratios (OR) and control for potential confounders. Results. Among PD patients, 44% had ever smoked, as compared to 59% in both control groups. Among ever-smoking patients, 74% quit prior to the date of diagnosis, as compared to roughly 45% of the ever-smoking control subjects. The OR for ever having smoked was 0.5 (95% confidence interval [CI] : 0.3-0.7), P trend < 0.00005). Conclusions. The results are considered in terms of criteria for causality. Plausible explanations for the observed inverse association between smoking and PD include: 1. A genetic predisposition that increases the risk for PD (such as defective detoxification enzymes) simultaneously decreases the likelihood of smoking. 2. Inherently lower dopamine levels in predestined PD patients cause them to be less prone to addiction. 3. Smoking is neuroprotective.