This study was designed to investigate the effects of incidental unfocused prompts and recasts on improving English as a foreign language (EFL) learners' grammatical accuracy as measured in students' oral interviews and the Test of English as a Foreign Language (TOEFL) grammar test. The design of the study was quasi-experimental with pre-tests, immediate post-tests and delayed post-tests. From three intact advanced EFL classes, three groups were formed: recast (n = 20), prompt (n = 20) and control (n = 20). Teachers in the two experimental groups responded to student grammatical errors with, respectively, recasts or prompts, while the grammatical errors of the control group were ignored and content feedback only was provided. The overall accuracy was measured by means of oral interview scores given by two trained raters and the grammar section of paper-based TOEFL. One-way repeated measures ANOVAs with post-hoc comparisons indicated that both prompt and recast groups significantly outperformed the control group, and the prompt group performed better than the recast group both in the immediate and delayed post-tests. Although attempts were made to make learners aware of the purpose of corrective feedback (CF), these findings show prompts which pushed learners to notice and self-correct their grammatical errors were more facilitative in improving their grammatical accuracy than recasts. Overall, these results suggest that the understanding, noticing, mental processing and active repair that prompting brings about are key factors in improving adult EFL learners' general oral accuracy. Implications for more effective unfocused prompts and recasts are also discussed.