This study investigated the differential effects of prompts and recasts, in the context of dyadic interaction, on the acquisition of grammatical gender by adult second language learners of French. Participants were 25 undergraduate students enrolled in an intermediate-level French course at an English-speaking university. All students were exposed in class to a 3-hr form-focused instructional treatment distributed over 2 weeks and were then randomly placed in either the recast or prompt group. On two occasions outside of class, individual students participated in three different oral tasks during dyadic interaction with a native or near-native speaker of French who, following learner errors in grammatical gender, provided feedback in the form of either prompts or recasts. Pretests and immediate and delayed posttests included two oral production tasks and a computerized reaction-time binary-choice test. Results of repeated-measures ANOVA showed that both groups significantly improved accuracy and reaction-time scores over time, irrespective of feedback type. We conclude that learners receiving recasts benefited from the repeated exposure to positive exemplars as well as from opportunities to infer negative evidence, whereas learners receiving prompts benefited from the repeated exposure to negative evidence as well as from opportunities to produce modified output.