Recent case-series studies indicated that a medication used to treat Parkinson's disease (PD), in particular Pramipexole, is associated with gambling. A case-series study cannot test this hypothesis; therefore, we need to design a case-control or cohort study to test the aforementioned hypothesis. Typical of a case-control design, we sampled on the dependent variable, which we defined as incident gambling in PD. A research neurologist, who was kept uninformed of the case-control status, retrospectively measured the exposure of interest (i.e. medications used to treat PD) by using the medical database system of Mayo Clinic Jacksonville. Eleven patients with PD without history of gambling, but had newly developed gambling, were matched by age and sex to the control group of 37 PD patients without gambling at a ratio of one case to at least three controls. Disease duration, age, and sex did not differ between cases and controls. Combined therapy with Pramipexole and levodopa did not increase the risk of gambling as compared to monotherapy with Pramipexole (OR, 0.15; 95% CI, 0.01-1.26). Treatment with Pramipexole was associated with increased risk of gambling and this association approached significance (OR, 3.6; 95% CI, 0.9-14.9). Patients with PD who newly developed gambling behavior were more likely to have been taking Pramipexole than other anti-PD medication. However, the association between Pramipexole and gambling behavior is not necessarily etiologic.