Clinical gait analysis is widely used by different therapists working with hemiplegic patients. The purpose of this study was to assess the reliability of video-based clinical gait assessment, as performed by practitioners in diverse specialties. Five neurologists, 5 physiotherapists, and 5 doctors of physical medicine and rehabilitation (physiatrists) were asked to study a videotape of 6 patients with hemiplegia in the act of walking. This activity was chosen due to the wide use of gait information for therapeutic prescription and medical decision-making during medical consultations, at conventions, or in exchanges between therapists. Results highlighted a certain inconsistency in the use of the gait description indicators. The 15 therapists used 396 different locutions to describe the gaits of the 6 patients. These locutions yielded 60 general indicators, or gait disorders, which were grouped in 5 categories. Eleven of them were classified as "controversial" due to the significant inter-subject variability of the evaluations. The study identified a large number of indicators that were used relatively inconsistently by the 3 specialties studied. The results of this study would appear to indicate that greater caution is needed when dealing with some of the "controversial" indicators, as well as with the "unusual" gait patterns observed in some patients.