An important, but as yet unresolved pedagogical question is whether discovery-oriented or direct instruction methods lead to greater learning and transfer. We address this issue in a study with 101 fourth and fifth grade students that contrasts two distinct instructional methods. One is a blend of discovery and direct instruction called Invent-then-Tell (IT), and the other is a version of direct instruction called Tell-then-Practice (TP). The relative effectiveness of these methods is compared in the context of learning a critical inquiry skill--the control-of-variables strategy. Previous research has demonstrated the success of IT over TP for teaching deep domain structures, while other research has demonstrated the superiority of direct instruction for teaching simple experimental design, a domain-general inquiry skill. In the present study, students in both conditions made equally large gains on an immediate assessment of their application and conceptual understanding of experimental design, and they also performed similarly on a test of far transfer. These results were fairly consistent across school populations with various levels of prior achievement and socioeconomic status. Findings suggest that broad claims about the relative effectiveness of these two distinct methods should be conditionalized by particular instructional contexts, such as the type of knowledge being taught.