Physiotherapists are required to record treatment time in both clinical research and daily practice. This study aimed to investigate the accuracy of treatment time records using a specific method of defining treatment for patients with stroke. Twenty-six physiotherapists provided time records using this method for one of their treatment sessions, each of which was videotaped. Two physiotherapists observed each videotape and their recorded times were then averaged to provide criterion scores against which clinicians' scores were compared. Clinicians' recordings were mostly of moderate to high accuracy and did not demonstrate systematic errors, despite the potential for accuracy to be compromised by many factors. This study has provided a model of how the accuracy of treatment time records can be validated.