OBJECTIVES: To develop a screening tool for timely referral for advanced therapy (AT) in patients with Parkinson disease (PD), and to compare the newly-developed tool with the published 5-2-1 criteria. DESIGN: Cross-sectional, diagnostic, observational study and multivariable logistic regression analysis for item selection. SETTING: 8 hospitals in the Northern part of the Netherlands situated in the catchment area of a specialized movement disorder centre. PARTICIPANTS: 259 consecutive PD patients not yet on AT visiting the outpatient clinic of participating hospitals from February 2017 to July 2018. PREDICTORS: 24 patient and disease characteristics as assessed by the treating neurologists, and scores on the NMS questionnaire. OUTCOME: Apparent eligibility for referral for AT based on consensus by a panel of 5 experts in the field of AT. RESULTS: 17 patients were deemed eligible for referral for AT (point prevalence: 6.6%). Presence of response fluctuations and troublesome dyskinesias were the strongest independent predictors of being eligible for referral. Both variables were included in the final model, as well as levodopa equivalent daily dose. Decision curve analysis showed that the new model outperformed the 5-2-1 criteria. A simple chart was constructed to provide guidance for referral. Discrimination of this simplified scoring system was good (AUC after bootstrapping: 0.97). CONCLUSION: The screening tool may improve efficiency of referral and subsequent treatment with AT in patients with PD. External validation is required prior to application in daily practice.