Problem-based learning (PBL) is widely endorsed as a desirable learning method, particularly in science. Especially in light of the method's heavy demand on resources, evidence-based practice is called for. Rigorous studies of the method's effectiveness, however, are scarce. In Study 1, college students enrolled in an elementary physics course participated in 7 hours of instruction on two concepts, one (electromagnetic field) via a PBL method and the other (gravitational field) via a lecture/discussion method. A comparison class enrolled in the same course with the same instructor underwent the identical experience except that matching of concept and method was reversed. Multiple assessments over time showed superior mastery of the concept learned via PBL for both groups. In Study 2, using a subtractive method, we further investigated the role of specific components of the PBL method by comparing a group who worked individually using a PBL method to one who participated in the typical collaborative form of PBL. Rather than the social component typically associated with PBL, the effective component of the method appears to be the focus on engagement with a problem. © 2010 Wiley Periodicals, Inc.