Multimodal feedback could promote college students' conceptual, procedural and heuristic knowledge learning. This study used an experimental design to investigate the interaction effects of instructor's narrated feedback (with procedural feedback delivered visually and heuristic feedback auditorily) and gender on their post-test performance of a similar problem. Ninety-three first-year undergraduate students participated in the study and were randomly assigned into different feedback groups. Students in the control group received visual-only feedback containing procedural and heuristic information from the instructor. Those in the treatment group viewed a generic three-minute video with the same procedural message presented visually and heuristic knowledge in narration. A two-way ANOVA showed that students receiving the multi-modal feedback scored significantly higher than those receiving visual-only feedback on the post-test. This effect did not differ for female or male students. The result supports that multimodal feedback could be a valuable tool to improve college students' mathematical problem solving skills. (Contains 6 tables and 2 figures.)