Spanish native speakers are known to pronounce onset /sC/ clusters in English with a prothetic vowel, as in "esport" for sport, due to their native language phonotactic constraints. We assessed whether accurate production of e.g. "spi" instead of "espi" was related to accurate perceptual discrimination of this contrast in second language (L2) speech of Spanish-English sequential bilinguals. A same-different discrimination task in stimulus pairs such as "spi"-"espi" assessed speech perception and a phonemic verbal fluency task elicited speech production. Logistic mixed model regressions revealed significant differences in accuracy between the bilinguals and the English monolinguals, although some bilinguals performed within the monolingual range. For the production task, but not for the perception task, bilinguals with more exposure to English and greater grammatical knowledge of English performed significantly more accurately than those with less exposure and lower grammatical knowledge. There was no significant correlation between production accuracy and perception accuracy. Through examining phonotactic constraints, these results expand a growing body of research into single sounds which suggests dissociations between L2 perception and production. In contrast to predictions made by L2 speech models, the findings indicate that accurate L2 perception is neither necessary nor sufficient for accurate L2 production, and instead are interpreted to indicate that the two capacities recruit different executive control mechanisms and are acquired--at least to a certain extent--independently in L2 acquisition.