This paper describes an educational improvement initiative. The purpose of this paper is to propose an evaluation for a course using Problem-based Learning with computer simulation (PBLCS). Since it is not the traditional way of teaching, therefore the evaluation should be different. The process in this study will be as follows: (1) review the most relevant papers in the related areas; (2) extract key evaluation criteria; (3) analyze the current course evaluation questions of two schools and, finally, (4) propose the new course evaluation for the PBLCS course. PBLCS is a teaching method that enables students to learn with their own experience of solving a problem in a computer. The relevant literature suggests that in learning with simulation, educators should consider and assess four learning outcomes. These are: REACTION (check how students subjectively feel and react to this particular learning method), LEARNING (check how much students have learned), BEHAVIOUR (check students’ post-learning competency) and RESULT (evaluate whether students utilize lessons and knowledge). Wongsurawat (2011) proposed that evaluation can reflect a better feedback if a student’s characteristic is reliable and well representative of the whole class answer. This can be done by calculating the correlation of student’s evaluation with the class average answer on close-ended subjective questions and the correlation of student’s evaluation with the class average answer on close-ended objective questions. If the close-ended answers from the student is reliable and well representative, a student’s answer in the open-ended questions should be more useful. It is expected that this proposed course evaluation should improve the reflection in different aspects on the quality of the content and material; technical support; instructor’s contribution; teaching and learning outcomes. With this rationale, a new set of evaluation for a course using PBLCS is proposed.