Recently, crime prevention is one of the overall strategies to reduce crime around the world including Malaysia. This is in lines with one of the Malaysian government agendas as stated in National Key Result Area (NKRA) which is to reduce crime. Because of the decision making process for crime prevention is manually done by the police, so it is difficult to determine the level of crime based on crime volume. That requires a method of selection to help the police to determine which is the most appropriate decision based on the level of crime. Therefore, this research developed prototype of decision support system (DSS) by using Mamdani-type fuzzy inference system (FIS) to classify the crime level and to improve the process of decision making for crime prevention. But, it is well-known that selection of membership functions (MF) is the key problem in the design of a FIS. Therefore, this paper outlines the basic comparisons between the triangular MF and trapezoidal MF. The experimental results of the two MFs are compared. It also shows which one is a better choice of the two MFs used in FIS in order to improve the efficiency of decision making for crime prevention. The result of the study indicates that triangular MF gives better classification of crime level but trapezoidal MF gives result closer to triangular MF.