The effectiveness of problem-based learning (PBL) in terms of increasing students' educational attainments has been extensively studied for higher education students and in nonexperimental settings. This paper tests the effectiveness of PBL as an alternative instruction method in secondary education. In a controlled experiment at the class level, we estimate its effect on tested student attainments, on perceived student attainments, on autonomous and controlled motivation, and on class atmosphere. The outcomes indicate a significant positive effect on student achievements, a significant positive effect on motivation, and a significant positive effect on class atmosphere.