Elicited imitation or sentence repetition tasks (SRT) are in common use in assessing linguistic (grammatical) ability, and to some extent for sign languages. An SRT is used to measure language proficiency and track progress over time. This paper reports on the creation and piloting of the first SRT for South African Sign Language (SASL), thus providing deaf schools with a language testing instrument, since there is currently no such instrument readily available. We specifically looked at the ability to perform the required task, lexical variation, and the appropriateness of particular grammatical structures of SASL. The SRT was created with 20 sentences, organised into three categories, according to estimated grammatical complexity of the sentences. Selection of grammatical structures was informed by SRTs developed for other sign languages. Data from 40 deaf children (7-9yrs, with a minimum of one year of exposure to SASL) were collected from two schools for the deaf in the Western Cape. The data was scored using absolute scoring, with a complete match being determined on the basis of word order, lexical items used, and non-manual features. Sentences scored as incorrect included sentences skipped and those with extra elements added (unanalysable). The percentages of correct responses showed an increase across the age range but were very low (2-18%). The length of exposure to SASL was a significant factor. Participants from one school performed significantly better on the test. Since they had fewer skipped and unanalysable utterances this may suggest greater familiarity with language testing. Despite previous checking of the lexicon, variation posed a challenge. While the categorisation of sentences was found to be relatively accurate, some adjustments are necessary for future use of the SRT in a broader context. Some grammatical structures were clearly too complex for the age groups tested. Sentence length also had an effect.