Educational system of Iran is suffering from the lack of active learning methods due to various reasons. Active learning methods motivate students to think and learn. One of these methods is collaborative learning which is more efficient than individual learning and develops a positive attitude in students toward teaching and learning. Another method is formative evaluation which enhances learning process. In this paper we investigate collaborative learning and formative evaluation in distinct groups. We use two hypotheses to address the difference between mathematics performance of students in conventional, collaborative learning, and formative evaluation methods. Adopting a semi-empirical approach and using multistage quasi-cluster sampling method, we chose 89 female students of the first year of secondary school in district 1 of Tehran in school year 2013-2014. We divided them into three groups of conventional, collaborative learning, and formative evaluation, and provided them with a researcher-made mathematics examination. ANOVA test results showed that students of formative evaluation group had better mathematics performance than conventional and collaborative learning groups. Therefore, formative evaluation seems to be an efficient method for teaching mathematics.