In view of problems of the fault leakage judgment, misjudgment and some faults which can not be judged in transformer fault diagnosis, this paper presents a transformer fault diagnosis model based on fuzzy Petri nets. The transformer fault diagnosis model is used to judge the type of the transformer faults through the analysis of the oil dissolved gas concentration. The method firstly uses gas chromatography to get the concentration value of each gas, and then calculates the rate of the gas growth to judge whether the faults happen. If the fault happens, the probability of each type of failure is calculated based on the model of the Petri nets. The method can avoid the lack of coding of three ratios, so that it can solve effectively the problems of the transformer fault diagnosis, misdiagnosis and missed judgment, as well as can provide the probability of each failure in a quantitative way, thus it improves the diagnostic accuracy.