The relationship between Parkinson disease (PD) and smoking has been examined in several studies, but little is known about smoking in conjunction with other behaviors and a family history of PD. Using unconditional logistic regression analysis, we studied individual and joint associations of these factors with idiopathic PD among 1,808 Danish patients who were diagnosed in 1996-2009 and matched to 1,876 randomly selected population controls. Although there was a downward trend in duration of smoking, this was not observed for daily tobacco consumption. A moderate intake of caffeine (3.1-5 cups/day) was associated with a lower odds ratio for PD (0.45, 95% confidence interval: 0.34, 0.62), as was a moderate intake of alcohol (3.1-7 units/week) (odds ratio = 0.60, 95% confidence interval: 0.58, 0.84); a higher daily intake did not reduce the odds further. When these behaviors were studied in combination with smoking, the odds ratios were lower than those for each one alone. Compared with never smokers with no family history of PD, never smokers who did have a family history had an odds ratio of 2.81 (95% confidence interval: 1.91, 4.13); for smokers with a family history, the odds ratio was 1.60 (95% confidence interval: 1.15, 2.23). In conclusion, duration of smoking seems to be more important than intensity in the relationship between smoking and idiopathic PD. The finding of lower risk estimates for smoking in combination with caffeine or alcohol requires further confirmation.