The present study examines the effects of pre-task explicit instruction on second language (L2) oral self-repair behaviour while controlling for the effects of working memory. The participants were 121 Iranian learners of English at incipient levels of language proficiency. Their working memory was measured using an operation span task and then they were randomly assigned to a control and an experimental group. Both groups performed a picture story-retelling task that was preceded by five minutes of planning time. During the planning time, the experimental group also received a grammar handout that explained English relative clauses. The instances of self-repairs were identified through stimulated recall interviews that immediately followed performance on the oral task. They were then classified into categories of global form repairs (F[subscript G]-repairs), local form repairs (F[subscript L]-repairs), and content repairs (C-repairs). A series of one-way ANCOVAs were run, the results of which indicated pre-task explicit instruction had significant and beneficial effects on F[subscript L]-repairs and adverse effects on C-repairs. The covariate was only associated with repairing the target structure. The findings are discussed in light of the Extended Trade-off Hypothesis and confirm the view that pre-task explicit instruction tends to foster a focus on form at the expense of meaning.