Purpose: To determine the psychometric properties of the Preschool Repetition (PSRep) Test (Roy & Chiat, 2004), to establish the range of performance in typically developing children and variables affecting this performance, and to compare the performance of clinically referred children. Method: The PSRep Test comprises 18 words and 18 phonologically matched nonwords systematically varied for length and prosodic structure. This test was administered to a typical sample of children aged 2;0-4;0 (n = 315) and a clinic sample of children aged 2;6-4;0 (n = 168), together with language assessments. Results: Performance in the typical sample was independent of gender and socioeconomic status but was affected by age, item length, and prosodic structure and was moderately correlated with receptive vocabulary. Performance in the clinic sample was significantly poorer but revealed similar effects of length and prosody and similar relations to language measures overall, with some notable exceptions. Test-retest reliability and interrater reliability were high. Conclusion: The PSRep Test is a viable and informative test. It differentiates within and between typical and clinic samples of children and reveals some unusual profiles within the clinic sample. These findings lay the foundations for a follow-up study of the clinic sample for investigation of the predictive value of the test.