BackgroundPeople with Parkinson's disease (PwPD) lose the ability in backward walking which is an important part of mobility in daily life. The 3-m backward walk test (3MBWT) evaluates backward walking; however, its reliability and validity have not been examined in PwPD yet.AimsTo examine (1) the test-retest reliability of the 3MBWT in PwPD; (2) the minimum detectable change in the 3MBWT times; (3) the concurrent and known-groups validity of the 3MBWT; and (4) the optimum cutoff time which best discriminates fallers from non-fallers with Parkinson's disease (PD).MethodsThis cross-sectional study included 36 PwPD and 33 healthy people. The 3MBWT was conducted with the 10-m walk test, timed up and go test, Berg Balance Scale, four square step test, activity-specific balance confidence scale, Movement Disorders Society Sponsored Unified Parkinson's Disease Rating Scale, and Hoehn and Yahr Scale.ResultsThe 3MBWT demonstrated excellent test-retest reliability (ICC = 0.965). The MDC of 2.13 s was determined. The 3MBWT had moderate to high correlations with the other outcome measures (correlation coefficient ranged from -0.592 to 0.858). On the 3MBWT times, there were significant differences between PwPD and healthy people, and between fallers and non-fallers with PD (p < 0.001 and p < 0.001, respectively). A 3MBWT time of 10.31 s was found to best discriminate fallers from non-fallers with PD.ConclusionsThe 3MBWT is a reliable, valid, and easy to administer outcome measure to assess backward walking performance in PwPD, indicating it to be used in practice and research.