The exponential growth of Portuguese-language legal documents has renewed interest in Question Answering (QA) systems capable of returning concise, legally sound answers to natural-language queries. This study presents a systematic literature review, conducted according to PRISMA 2020 guidelines, that synthesises current evidence on QA techniques applied to Lusophone legal texts. Searches, without temporal restrictions, were executed in nine databases (ACM, El Compendex, ISI Web of Science, Periódico Capes, Scielo, Science@Direct, Scopus, Sol SBC and Springer Link) using a string that combine jurisprudential, linguistic and methodological terms. After duplicate removal, independent screening and quality appraisal, ten primary studies met the inclusion criteria (peer-reviewed publications developing or evaluating QA pipelines over Brazilian or Portuguese legislation). Publication activity is recent: more than 70% of the papers appeared between 2023 and 2025 and focus on Brazilian statutes and court decisions. Most pipelines adopt hybrid retrieval—BM25 or symbolic regex filters coupled with BERT-family dense encoders fine-tuned on legal corpora, while Retrieval-Augmented Generation with GPT-class models emerges in the latest research. Reported exact-match scores range from 0.60 to 0.83 and F1 from 0.75 to 0.87; however, only a quarter of the studies release code or data, hindering reproducibility. Common gaps include limited handling of the temporal validity of norms, scarce evaluation by legal specialists, and the absence of benchmark datasets for Portuguese. Overall, QA research for Lusophone law is accelerating yet remains fragmented; future work should prioritize shared resources, temporally aware models, and metrics that capture legal soundness beyond lexical overlap.