While previous research has indicated that learners with field-dependence (FD) and field-independence (FI) cognitive styles benefit differentially from different instructional modes, previous corrective feedback studies have ignored the issue of matching error correction strategies to learners' cognitive style. To shed some light on this issue, the current study examined the hypothesis that learners with FI and FD cognitive styles benefit differently from recasts. Seventy-six intermediate level EFL learners in four intact classes constituted the experimental and control groups, which differed from each other in terms of the presence or absence of recasts for their errors during classroom interactions. The participants were also classified as FD or FI based on their scores on the Group Embedded Figures Test (GEFT). Thus, four groups were formed: recasts-FD, recasts-FI, control-FD, and control-FI. The results of a writing task and a picture description task administered as pretest, posttest and delayed posttest indicated that the recasts-FI group significantly outperformed the other three groups in both posttests and delayed posttests, while no statistical difference was observed among the other three groups. In other words, as the results showed, only FI learners benefited from recasts.