Abstract Previous studies have shown that bilingual children typically score more poorly on nonword repetition (NWR) tasks than monolingual peers, which has been attributed to bilinguals’ lower proficiency in the language that the NWR task is based on. To enable fairer assessments of bilingual children, Cross-Linguistic NWR tasks (CL-NWR tasks) have been developed that are based on the linguistic properties of many languages. The aim of this study is to investigate whether young children’s performance on a CL-NWR is less dependent on existing knowledge of a specific language than performance on a Language-Specific (Dutch-based) NWR (LS-NWR). Bilingual and multilingual two- and three-year-olds ( N  = 216) completed a CL-NWR and LS-NWR, as well as a Dutch receptive vocabulary task. Parents reported the number of languages children spoke other than Dutch. Results of linear mixed-effect regressions showed that Dutch vocabulary scores related to performance on the CL-NWR task less strongly than to performance on the LS-NWR task. The number of non-Dutch languages spoken did not differentially relate to performance on the two tasks. These findings indicate that CL-NWR tasks – at least as used here – allow for more language-neutral NWR assessments within linguistically diverse samples, already at toddler age.