Purpose This study aimed to assess a protocol for the evaluation of developmental language disorder (DLD) in language minority bilingual children (LMBC). The specific aims were (a) to test group differences, (b) to evaluate the discriminant validity of single measures included in the protocol, and (c) to define which model of combined variables had the best results in terms of efficacy and efficiency. Method Two groups of LMBC were involved, one with typical development (n = 35) selected from mainstream schools and one with DLD (n = 20). The study protocol included the collection of demographic information and linguistic history; a battery of standardized tests in their second language (Italian), including nonword repetition, morphosyntactic comprehension and production, and vocabulary and narrative skills; and direct (children's evaluation) and indirect (parents' questionnaire) assessment of linguistic skills in their first language. Results Results showed that the two groups differed in almost all linguistic measures. None of the single measures reached good specificity/sensitivity scores. A combined model that included direct and indirect assessment of first language skills, morphosyntactic comprehension and production, and nonword repetition reached good discriminant validity, with 94.5% of cases correctly classified. Discussion The study defines a complex picture of the linguistic profile in bilingual children with DLD, compared to typically developing bilingual peers. The results reinforce the idea that no single measure can be considered optimal in distinguishing children with DLD from typical peers. The study offers a concrete example of an effective and efficient protocol with which to discriminate LMBC with and without DLD.