This study aimed to evaluate the effectiveness of cooperative learning and direct instruction in improving floor gymnastics skills among physical education students. A quantitative approach employing a quasi-experimental pretest–posttest control group design was used. Eighty students participated in the study and were assigned to either the Cooperative Learning group (n = 40) or the Direct Instruction group (n = 40). Floor gymnastics performance was assessed using a validated performance rubric covering four movements: Roll Backward to Handstand (RBH), Forward Handspring (FH), Front Balance (FB), and Shoulder Stand (SS). Performance was evaluated on a five-point rating scale by trained assessors, and the instrument's content validity was established through expert review by gymnastics specialists. Data were analyzed using the Wilcoxon Signed-ranks Test and the Mann–Whitney U Test. The results demonstrated significant improvements in all four floor gymnastics skills within both instructional groups (p < 0.05). Furthermore, the between-group analysis revealed that the cooperative learning group achieved significantly higher posttest performance than the direct instruction group across all four movements, with small to moderate effect sizes (p < 0.05). These findings indicate that although both instructional strategies effectively improved students' floor gymnastics performance, cooperative learning provided greater overall learning gains. Therefore, cooperative learning may be considered a more effective instructional approach for enhancing floor gymnastics skills while promoting active student participation in physical education.