Sign-language assessment has been an issue for 15 years. Several tools in different sign languages have been designed, and each of them has a specific goal: assessment of sign language as a first language (L1), sign language as a L2, lexical development, cognitive abilities, etc. (Haug, 2008. http://www.signlang-assessment.info). There is currently no available test to assess LSF. As a consequence, researchers, teachers and therapists have no reliable benchmarks for LSF acquisition, and do not know how to efficiently assess linguistic skills. In the wake of Courtin et al.’s work on LSF assessment (2010), and other sign-language assessment tools, our general goal is to develop a series of tests to assess linguistic abilities in LSF. Among the numerous tasks that allow us to evaluate language skills, the Sentence Repetition Task (SRT) seems quite relevant for sign languages. The SRT is often used in Vocal Languages (VL). It results in a good representation of language abilities while being quick to run and easy to score. This task is successful with adults, but children tend to fail because their phonological skills and linguistic representations are not strong enough (Mason et al, 2010). However, it is also considered as a relatively reliable marker of language development and language processing (Chiat et al., 2013). In addition, a poor performance on the task is considered to be a typical marker of Specific Language Impairment (for VL, Conti-Ramsden et al., 2001; for SL, see Marshall et al., 2015). The aim of the present study is twofold. First, we will present a new Sentence Repetition Task elaborated in LSF; second, we will study the role of Age of Acquisition (AoA), Length of Exposure (LoE), and developmental tendency (Chronologic Age) on sign language skills. For this purpose, sixty-two deaf children, aged from 6;1 to 12;8 were tested on their repetition abilities. Thirty-four children were native signers (i.e., exposed to LSF at birth), and twenty-eight children were late signers (i.e., exposed to LSF at school). All the children have no history of language disorders. The children were instructed to repeat 20 signed sentences from a video produced by a French native adult signer. In order to quantify lexical, phonological and morphosyntactic errors, we compared the children’s repetitions to the production of the native signer in the video. As expected, our data showed better repetition ability in the older children, regardless of AoA. However, for younger children, and even more when they are late signers, while their repetition performances were quite good, their repetition skills proved significantly different: more sign omissions, unfinished sentences, inaccurate phonological ability and errors in morphosyntactic localisations. These data constitute a first step in the process of providing a normative database. Further investigations will aim to combine these repetition data with narrative production of a cartoon (Caet & Blondel, 2018). Another interesting perspective will be to test children with atypical acquisition such as late signers and children with Specific Language Impairment (work in progress).