Background and purpose Anxiety is potentially a pre‐motor symptom of Parkinson's disease (PD). Our aim was to investigate the association between anxiety and subsequent PD risk in a population‐based sample. Methods A total of 174 776 participants, who were free of prior PD, dementia and stroke, were enrolled from Taiwan National Health Insurance Research Database between 1 January 2005 and 31 December 2005. The association between anxiety at the beginning of the study and the incidence of PD was examined using a Cox regression model. Information regarding comorbidities, especially depression, and concomitant medication use was adjusted in the proportional hazards models. Results Over an average follow‐up of 5.5 years, 2258 incident PD cases were diagnosed. After adjusting for age, sex, comorbidities and concomitant medication use, patients with anxiety were more likely to develop PD than subjects without anxiety adjusted hazard ratio (HR) 1.38; 95% confidence interval (CI) 1.26–1.51. Anxiety severity was dose‐dependently associated with increased likelihood of PD: crude HR 1.27 (95% CI 1.11–1.44) for mild anxiety, 1.35 (95% CI 1.19–1.53) for moderate anxiety and 2.36 (95% CI 2.13–2.62) for severe anxiety (P < 0.0001). Results were similar after adjustment for age, sex, comorbid depression and other PD risk factors, and in the sensitivity analyses excluding participants with comorbid depression or with a PD diagnosis <3 years after anxiety diagnosis, and controlling for Charlson's scores. Conclusions The likelihood of developing PD was greater amongst patients with anxiety than patients without anxiety, and the severity of anxiety correlated with risk of PD.