There has been a growing interest in the examination of the steady state of simultaneous bilinguals. An understanding of what leads to the possible weaknesses in the grammar of early bilinguals can contribute to our understanding of the possible causes of the apparent characteristic ‘failures’ in second language acquisition (Montrul 2008). Spanish has a gender feature for nouns (Carroll 1989) and gender agreement for determiners and adjectives. Problems with the acquisition of gender marking on the noun and/or with gender agreement are well-known in the L2 literature (Hawkins 1998; Fernández–Garcia 1999; Franceschina 2001; Bruhn de Garavito and White 2002; White et al. 2004). The question is whether or not the gender feature is acquired on the noun itself (White et al. 2004) or if problems are due to mapping. Current research has suggested masculine as the default gender for agreement marking in Spanish (McCarthy 2007) but the question of whether the gender feature is acquired on the noun persists. In the present study we will focus on the status of an internal interface property, gender, where syntax interfaces with morphology by examining code-mixed DPs of Spanish/ English bilinguals. A salient characteristic of bilingual grammars is that of code-mixing where properties from two languages are systematically ‘mixed’ within a phrase or sentence (Toribio 2001). The treatment of code-mixed DPs is different for L1 Spanish speakers who speak English than for bilingual L1 speakers of Spanish English (Liceras et al. 2008). The Liceras et al. spontaneous code-mixing data has shown that in situations where the determiner is in Spanish and the noun is in English, a Spanish speaker will agree the determiner with the inherent gender of the noun (as in (1a-b)) while a bilingual speaker will invariably use the masculine form of the determiner regardless of the gender of the noun in Spanish (as in (2a-b): This evidence from the code-mixed DPs of Spanish/English bilingual speakers indicates that the status of their gender feature is different from that of L1 Spanish speakers. Our research questions are the following: Are differences in Det selection between bilingual and L1 Spanish speakers due to a difference in underlying structure (English DP versus Spanish DP)? Have bilinguals acquired the abstract gender feature of Spanish? In order to investigate the research questions we have developed two tests: a lexical selection task and an elicited production task. Both tasks target code-mixed DPs where the Det is provided by Spanish and the noun is in English. Target tokens look at the gender selection of the Det as well as agreement in copula constructions. There are three subject groups: simultaneous Spanish/English bilinguals, adult L2 Spanish advanced learners and L1 Spanish speakers. Data will be discussed in terms of the status of the gender feature, the issue of default word marking, and the acquisition of internal interface properties.