We conducted a systematic review and meta‐analysis aimed at establishing robust prevalence estimates and identifying clinical correlates of fatigue in PD. From 2,459 titles and abstracts, we selected 44 relevant studies (n = 7427 patients). Overall, the meta‐analysis showed a prevalence of fatigue of 50% in PD. This prevalence estimate, however, was significantly moderated by study heterogeneity in measurement scales and cut‐off thresholds. In contrast, demographic features, disease severity, cognitive impairment, and depression did not moderate prevalence estimates. Moreover, fatigue prevalence did not differ between de novo and treated PD patients. Compared to nonfatigued patients, fatigued patients had sligthly higher age (1.44 years), disease duration (0.93 years), l‐dopa equivalent daily dose (50.89 units), UPDRS‐III (4.99 points), and H & Y (0.33 points), as well as risk of comorbid depression (risk ratio = 1.89) and had a little lower MMSE score (–0.66 points). Fatigue was moderately associated with apathy (Hedges' g = 0.55), anxiety (Hedges' g = 0.67), daytime somnolence (Hedges' g = 0.43), sleep disturbances (Hedges' g = 0.66), and poorer quality of life (Hedges' g = 1.23). Our analyses suggest that fatigue is a frequent, independent nonmotor symptom in PD appearing early and persisting throughout the disease course, and that establishing uniform diagnostic criteria for PD‐related fatigue is critical. In addition, several nonmotor symptoms appear to be associated with fatigue and negatively impact quality of life. Pharmacological and nonpharmacological interventions targeting fatigue and associated symptoms may improve quality of life in patients with PD. © 2018 International Parkinson and Movement Disorder Society