Montessori classrooms vary in the degree to which they adhere to Maria Montessori's model, including in the provision of materials. Specifically, some classrooms use only Montessori materials, whereas others supplement the Montessori materials with commercially available materials like puzzles and games. A prior study suggested such supplementation might explain observed differences across studies and classrooms (Lillard, 2012), but an experimental study is the best test of this possibility. Here we present such an experiment, with 52 children in three Montessori classrooms, all of which included supplementary materials at the start of the study. Non-Montessori materials were then removed from two of the classrooms, and all children were given six pretests as a baseline. Four months later, children were retested to see how much they had changed across that period. Children in the classrooms from which the non-Montessori materials were removed advanced significantly more in early reading and executive function and advanced to some degree more in early math than children in the other two classrooms. There were no differences across the classroom types in amount of change on the tests of vocabulary, social knowledge, or social problem-solving skills.