The invention discloses a fault diagnosis method and system for an electric driven traffic tool. The method comprises the steps of obtaining running parameter data of a battery of a monitored traffic tool, and performing screening arrangement on the running parameter data to obtain a data matrix of voltages of battery cells; adopting a 3 sigma multi-level screening algorithm for the data matrix to obtain a fault quantity of each battery cell; calculating a fault probability of each battery cell according to the fault quantity; and judging whether the fault probability of each battery cell exceeds a second given value or not, and if yes, judging that the fault occurs. According to the method and the system, a probability characteristic of Gaussian distribution is combined with a 3 sigma confidence interval to form 3 sigma multi-level screening; for low-order fault data, all data exceeding the 3 sigma confidence interval can be removed by one-time screening; and multi-order fault data or fault data with relatively large order-of-magnitude difference can be screened for multiple times and removed to obtain a central value closest to real data, so that the processing efficiency is high.