Visser M, Marinus J, Bloem BR, Kisjes H, van den Berg BM, van Hilten JJ. Clinical tests for the evaluation of postural instability in patients with Parkinson’s disease. Arch Phys Med Rehabil 2003;84:1669–74. Objective To determine which test for postural instability in Parkinson’s disease (PD) is reliable, valid, and easy to perform in a clinical setting. Design Cross-sectional reliability and validity study. Setting Academic center for movement disorders. Participants Forty-two patients with PD and 15 controls. Based on the results of a structured interview, the patients were divided in PD-unstable (n=22) and PD-stable (n=20) groups. Interventions Not applicable. Main outcome measures Several variants of the retropulsion test with differences in execution and scoring. Responses were scored on 5 different rating scales (ratings of Nutt, Bloem, Pastor; the Unified Parkinson’s Disease Rating Scale [UPDRS]; the Short Parkinson Evaluation Scale). These tests were compared with steady-stance positions. Results The interrater reliability was high for most ratings, with weighted κ ranging from .63 for the UPDRS to .98 for both the Pastor rating and steady-stance positions. Most ratings distinguished between the groups. However, the Nutt rating had the highest overall predictive accuracy, with a sensitivity of .63 and a specificity of .88. Conclusions The most valid test for postural stability in PD was an unexpected shoulder pull, executed once, with taking more than 2 steps backward considered abnormal. This retropulsion test is easy to use in a clinical setting.