To improve the traffic safety in the rural-urban continuum, it is necessary to be identified the parameters and made the evaluation of accident parameters accurately and scientifically. Firstly, the accident parameters were extracted from the accident data of the rural-urban continuum in Yunnan Province and the accident type was divided into death accident and injury accident. Secondly, the extracted accident parameters were coded. In binary logistic regression analysis, principal component analysis was introduced to build the prediction model of traffic accident death of the rural-urban continuum, and the influence of parameters on the occurrence of death accidents were analyzed. Compared with the traditional logistic model, the prediction accuracy of the model was improved by 9.7%, and the number of iterations is reduced by 244 times. The results show that driver"s age of, time and vehicle type are the main factors of the death accidents in the urban-rural continuum. When the driver"s age is 20-30 years old, the time is night, and the vehicle type is minivan or electric vehicle, the probability of the death accident is the highest in the urban-rural continuum. The model can objectively and accurately predict and evaluate the death traffic accidents in the rural-urban continuum.