The　stability　of　time　series　and　the　fatality　rate　of　100,　000　population　of　traffic　accidents　in　Beijing　are　analyzed　by　ARIMA　model.　The　ARIMA　model　forecasted　by　spss　11.5　software　and　it　improved　the　precision　of　the　forecasting　result.　The　ARIMA　model　is　able　to　be　used　for　all　kinds　of　seasonal　and　non-seasonal　time　series　in　practice.　The　forecasting　data　obtained　from　the　ARIMA　model　simulates　the　actual　data　well.　The　result　is　very　important　to　the　government　for　countermeasures　making　which　help　prevent　traffic　accidents　from　happening.