The LITMUS-QU-NWR varies the nonwords by syllabic complexity (i.e., consonant clusters and word-internal codas (dos Santos & Ferré, 2018; Grimm, 2022). This study aims to shed light on the production patterns of word-initial consonant clusters in order to verify that BiDLD children struggle with phonological complexity (de Almeida et al., 2019). Based on the LITMUS-QU-NWR, we analyzed how German- and French-speaking BiDLD children produced word-initial obstruent+liquid (OL; /fl/, /kl/, /pl/) and s+obstruent clusters (sC; /sk/, /sp/, /spl/, /skl/) compared to their BiTD and MoTD peers. Two questions were addressed: Do the BiDLD children differ from BiTD and MoTD children in overall accuracy of word-initial clusters? Given the construction principles, we expected effects of clinical status (BiTD vs. BiDLD) but no effects of language status (MoTD vs. BiTD). Second, do the groups differ regarding the reduction patterns of consonant clusters? Based on the literature (cf., Gerrits, 2010 for Dutch), we predicted preservation of the less sonorant, i.e. of C1 in OL and of C2 in sC-clusters in all groups. 93 French-speaking (age range: 64-107 months) and 147 German-speaking children (age range: 65-95 months) participated in the study. LMM analyses showed significant effects of clinical status but not of bilingualism in German in word-initial OL- and sC-clusters, all p < .017), in line with our predictions. This was also the case for sC-clusters in French (all p < .001). Bilingualism and clinical status mattered for OL clusters in French (MoTD children produced more OL-clusters than BiTD and BiDLD children, all p <.021). Due to the older age, the French BiDLD children performed better than the German BiDLD group with regard to sC-clusters. As predicted, all groups had a preference to maintain C1 in OL and C2 in sC clusters. The results provide further evidence that the production of word-initial consonant clusters is a valid indicator of DLD in bilingual children (de Almeida et al., 2019). We will discuss how factors like the phonological properties of the target language, the structure of the L1, the child’s language exposure, and the severity of the impairment in the DLD group can account for our findings.