A fuzzy model is developed to address traffic behavior in determination of traffic signal time at intersection. The proposed fuzzy model involves fuzzification of data, fuzzy inference engine, and defuzzification of the output of the inference engine. One of the main benefit of fuzzy model, is that the model is able to involve linguistic parameters, as well as quantitative ones. It is one distinctive character of fuzzy model that the system may involve uncertainties or ambiguities coming from human language. The proposed model was tested against real data coming from data acquisition from several samples of intersections in Bandung, during the period of May-June 2017. The behavioral aspects included in term model were types of violation such as vehicles stopping and blocking the movement left turn on red or vehicles stop at a special stop room motor cycles. Improvement of intersection performance is indicated by decrease of the length of vehicle queues, and the increase of intersection capacity. Analysis of the proposed method shows an average of 56% extension of green time over existing condition therefore the proposed method offers contribution to a better regulation of traffic signals at isolated intersections. Similar approach can be further developed for other issues of traffic signals, that matches the criteria of isolated intersections, such as the Jl.Pasteur-Jl.Pasirkaliki intersection, the Jl.Cibaduyut-Jl.Soekarno Hatta intersection, the Jl.Kiaracondong-Jl.Soekarno Hatta intersection and the Jl.Gedebage-Jl.Soekarno Hatta intersection.