This research aims to examine differences in students' understanding of accounting taught with Blended Learning, Jigsaw, and Direct Instruction learning models. The research was conducted at the University of ABC. The study population numbered 264 students, then the sampling technique used cluster random sampling, and the number of students being sampled was 60. This type of research is quasi-experimental. Data collection techniques using learning outcomes tests. The data analysis technique used one ANOVA. The results showed (1) There was a significant difference in the understanding of accounting taught by the Blended Learning model, Jigsaw, and Direct Instruction, (2) Students who were taught with Blended Learning had a significantly higher understanding of accounting than students taught with Jigsaw, (3) Students who were taught with Blended Learning had a higher understanding of accounting than students who were taught with Direct Instruction, (4) Students who were taught with Jigsaw had a lower understanding of accounting than students who were taught with Direct Instruction. In conclusion, Blended Learning, Jigsaw, and Direct Instruction learning models positively affect understanding accounting. The most effective learning model in this research was Blended Learning. Blended Learning, Jigsaw, and Direct Instruction learning models can be used to teach accounting; however, to maximize the effect of Blended Learning, Blended Learning requires a powerful internet connection.