This paper proposes guidelines to evaluate agent’s internal model effectively in organizational learning. We developed an approach named Agent-based Organizational Cybernetics (AOC) to describe problems in organizational learning with an agent-based model and to design some prescriptions for them. This paper focuses on four typical evaluation problems of internal model in organizational learning: (1) environmental scanning, (2) policy making, (3) coordination, and (4) task resolution. We design and test some guidelines for each problem. The results provide some effective guidelines in organizational learning. Organizational-performance(OP)-based evaluation realizes organization’s high performance in all four problem situations. However, the OP-based evaluation cannot guarantee that an agent gets a right internal model all the time. So, it is effective to also consider some other complementary evaluations.