In this talk I will discuss data from a study which investigated the role of syntactic processing in driving bilingual language selection. In two experiments, 120 English-dominant Spanish-English bilinguals read aloud 18 paragraphs with language switches. In Experiment 1a, each paragraph included eight switch words on function targets (four that repeated in every paragraph), and Experiment 1b was a replication with eight additional switches on content words in each paragraph. Both experiments had three conditions: (a) normal, (b) nouns-swapped (in which nouns within consecutive sentences were swapped), and (c) random (in which words in each sentence were reordered randomly). In both experiments bilinguals produced intrusion errors, automatically translating language switch words by mistake, especially on function words (e.g., saying the day and stay awake instead of the day y stay awake). Intrusion rates did not vary across experiments even though switch rate was doubled in Experiment 1b relative to Experiment 1a. Bilinguals produced the most intrusions in normal paragraphs, slightly but significantly fewer intrusions in nouns-swapped paragraphs and there was a dramatic drop in intrusion rates in the random condition, even though the random condition elicited the most within-language errors. Bilinguals also demonstrated a common signature of inhibitory control in the form of reversed language dominance effects, which did not vary significantly across paragraph types. Finally, intrusions increased with switch word predictability (surprisal), but significant differences between conditions remained when controlling for predictability. These results demonstrate that bilingual language selection is driven by syntactic processing, which operates independently from other language control mechanisms, such as inhibition.