We conduct quantitative analyses of 5,569 tokens of Spanish word-final /s/ in the speech of Spanish/English bilinguals and monolinguals to test the Exemplar Model of Lexical Representation. We establish that reduction (aspiration, deletion) of word-final /s/ in the speech of New Mexico is significantly lower in Spanish words whose English translation equivalent has a word-final /s/ and significantly higher in words with no word-final sibilant in their English counterpart. The same crosslinguistic similarity effect is not found in a nearly identical (monolingual) variety of Spanish (Chihuahua), owing, we argue, to the lack of influence from word-final /s/ in English. Results of variable rule analyses reveal that the bilingual exemplar connections are not uniform across all words. Rather, the crosslinguistic ties are strongest where more phonological, orthographic, or semantic overlap can be posited (i.e., cognates). Results provide evidence of bilingual lexical connections in support of an exemplar model of lexical representation.