Upon the latest reform to the college entrance examination in China (i.e., Gaokao), high schools began implementing an optional class system. Under this scheme, students’ time slots become complex, thereby increasing the difficulty in formulating a suitable timetable from the available ones. To address this problem, the course-scheduling model was improved. On the basis of the original hard constraints, the “concurrent group” was considered, and the softer constraints were regarded as optimization goals, such as “teaching plans synchronously”, “no idle periods in the timetables of teachers”, and “evenly distributed lessons”. Given these soft constraints, the model becomes more practical. In this study, a two-phase tabu search algorithm was proposed to solve the problem. The proposed algorithm uses the characteristics of the graph coloring model to eliminate redundant calculations in the neighborhood search process, thereby effectively improving computational efficiency. Fifteen practical instances of different scales were selected for testing to verify the effectiveness of the algorithm. The proposed algorithm can formulate high-quality available timetables (The average satisfaction rate of soft constraints is more than 71%) in a short period.