Prior research showed that interleaved practice (studying multiple skills at once) is more effective than blocked practice (studying only one skill at a time). This study aims to replicate the benefits of interleaved practice on the proceduralization of second language (L2) syntax and further examines the role of working memory (WM) in different practice schedules. Sixty English learners studied five types of relative-clause constructions under either blocked- or interleaved-practice conditions. The blocked-practice group engaged in systematic form-focused speaking practice with exemplars blocked by syntactic category, while the interleaved-practice group received mixed exemplars from the different categories. The proceduralization of grammatical knowledge was measured by analysing the accuracy and speed indices from a picture description test, which was administered immediately and one week after the training session. Learners' WM capacity was measured using a listening-span task. Results showed that interleaved practice led to more accurate performance on both immediate and delayed posttests than blocked practice. The advantage of interleaved practice was less pronounced for the speed dimension of performance. Furthermore, interleaved practice facilitated skill development regardless of learners' WM capacity, whereas in the blocked-practice condition, learners with higher WM capacity benefited more than those with lower WM capacity in speeding up of relative clause use, which presumably reflects the proceduralization--automatization stage.