This paper proposes a new model for team building, which enables teachers to build coherent teams rapidly and fairly for the term projects of software engineering courses. Moreover, the model can also be used to build teams for any type of project, if the team member candidates are students, or if they are inexperienced on a certain subject. The proposed model takes students' preferences and the teacher's considerations into account when a team building process is required for any type of course. In addition, this paper investigates how team building models (RandomM: teams are built with randomly selected students; TeacherM: teacher selects the members for each team; StudentsM: students build their own teams and the proposed model) affect team performance and how gender differences affect project activities and team performance. A three-year (five semesters) teaching experiment was performed with the participation of 248 male and 79 female university students and a total of 67 software project teams. Two different One-way ANOVA tests were applied on the experimental data, and the results indicated that the proposed model was better than RandomM, TeacherM and StudentsM models in terms of project grades, and the effect of gender differences on the teams' performance and project activities was negligible. (Contains 4 tables and 3 figures.)