Deaf communities display high level of variability in language proficiency (NIEDLE et al, 2000), which makes it difficult to structure samples for studies about the Grammar (I-language) in sign languages (SL). Such variability is directly related to the age of acquisition, the quality and quantity of the linguistic input received by deaf children during the acquisition period. Deaf signers whose parents are also deaf signers (DsDsP) are considered to be native signers. That is, these signers are considered to have full grammatical competence (Chomsky, 1965, 1986) on their native languages. However, due to shortage of DsDsP informants, researchers tend to leave open the set of signers with full grammatical knowledge, including deaf signers with non-deaf parents (DsNonDsP) as informants in their research. Studies suggest, however, the existence of a sensitive period for the acquisition of the structural properties of any given grammar. This sensitivity tends to be gradually reduced until the onset of puberty (Lenneberg, 1967). Therefore, we should, by hypothesis, expect that DsNonDsP with late exposure to any SL, including Brazilian Sign Language (Libras), do not have the same grammatical competence as DsDsP. Investigating this hypothesis is the main goal of our research, a theoretical-experimental study on grammatical competence in Libras. We focused on the morphosyntax of verbal agreement in Libras and we conducted an acceptability judgment take, in which signers (20 DsDsP and 45 DsNonDsP (15 with exposure to Libras before 4 years; 15 with exposure between 5 and 7 years; 15 with exposure from 8 years old)) used a 5-point Likert Scale (1 = Not acceptable, 5 = fully acceptable) to judge 32 target sentences/videos. The independent variables were: type of verbal agreement (regular vs. backwards), auxiliary verb (presence vs. absence) and morphological manifestation of agreement on the verb (partial (with the object only) vs. neutral (no agreement at all)). The data collected were statistically treated (ANOVA with repeated measures) and the results were analyzed taking into consideration the literature on language acquisition within Generative Grammar and the literature on verbal agreement in SL. The results did not show any strikingly difference between DsDsP and DsNonDsP. Both DsDsP and DsNonDSP seem to be sensitive to the type of verbal agreement, displaying a preference for regular verbal agreement and a lower preference for the presence of auxiliary verb even in sentences with neutral verbal agreement, in accordance with Lourenço (2014). However, DsDsP showed less acceptance of partial regular verbal agreement in the presence of auxiliary in accordance with Quadros and Quer (2008, 2010).