The authors in a convergent parallel mixed method design aimed to investigate whether and how the geometry knowledge and attitude toward geometry of students with field-dependent/independent cognitive style improve differentially within a project-based learning environment. Quantitative data were collected from 97 seventh-grade students before and after the treatment. Repeated measures multivariate analysis of covariance results, controlling for gender and prior mathematics achievement, revealed a statistically significant main effect of time but not cognitive style. Follow-up repeated measures analysis of covariance by controlling the experiment-wise alpha level indicated that field-dependent, field-mixed, and field-independent students achieved a statistically significant mean increase for all measures. The qualitative (classroom observations and interviews with some students) data provided evidence that contextualizing, visualizing, and collaborating opportunities in the project-based learning environment made contributions for mean increase.