Objective: To discover how reliably speech and language therapists could diagnose apraxia of speech using their clinical judgement, by measuring whether they were consistent (intra-rater reliability), and whether their diagnoses agreed (inter-rater reliability). Design: Video clips of people with communication difficulties following stroke were rated by four speech and language therapists who were given no definition of apraxia of speech, no training, and no opportunity for conferring. Settings: Videos were made of people following stroke in their homes. Ratings of the videos were carried out in the university lab under controlled conditions. Subjects: Forty-two people with communication difficulties such as aphasia, apraxia of speech and dysarthria took part, and four specialist speech and language therapists acted as raters. Main measure: Speech and language therapists' ratings of the presence and severity of apraxia of speech using videos. Results: Intra-rater reliability was high for diagnosing (1) the presence of apraxia of speech (Cohen's kappas ranging from 0.90 to 1.00; 0.93 overall), and (2) the severity of apraxia of speech (kappa 0.84 to 0.92; 0.90 overall). The inter-rater reliability was also high for both the presence of apraxia of speech (kappa 0.86) and severity of apraxia of speech (0.74). Conclusion: Despite controversy over its nature and existence, specialist speech and language therapists show high levels of agreement on the diagnosis of apraxia of speech using their clinical judgement. © Sage Publications 2007.