Numerous scales assess dyskinesia in Parkinson's disease (PD), variably focusing on anatomical distribution, phenomenology, time, severity, and disability. No study has compared these scales and their relative ability to detect change related to an established treatment. We conducted a randomized placebo-controlled trial of amantadine, assessing dyskinesia at baseline and at 4 and 8 weeks using the following scales: Unified Dyskinesia Rating Scale (UDysRS), Lang-Fahn Activities of Daily Living Dyskinesia Rating Scale (LF), 26-Item Parkinson's Disease Dyskinesia scale (PDD-26), patient diaries, modified Abnormal Involuntary Movements Scale (AIMS), Rush Dyskinesia Rating Scale (RDRS), dyskinesia items from the Movement Disorder Society-sponsored revision of the Unified Parkinson's Disease Rating Scale (MDS-UPDRS), and Clinical Global Impression (severity and change: CGI-S, CGI-C). Scale order was randomized at each visit, but raters were aware of each scale as it was administered. Sensitivity to treatment was assessed using effect size. Sixty-one randomized dyskinetic PD subjects (31 amantadine, 30 placebo) completed the study. Four of the 8 scales (CGI-C, LF, PDD-26, and UDysRS) detected a significant treatment. The UDysRS Total Score showed the highest effect size (η2 = 0.138) for detecting treatment-related change, with all other scales having effect sizes < 0.1. No scale was resistant to placebo effects. This study resolves 2 major issues useful for future testing of new antidyskinesia treatments: among tested scales, the UDysRS, having both subjective and objective dyskinesia ratings, is superior for detecting treatment effects; and the magnitude of the UDysRS effect size from amantadine sets a clear standard for comparison for new agents. © 2012 Movement Disorder Society.