The main purpose of the this essay has been to make use of one or several time series models in order to explain the correlation that is assumed to exist between the observed and estimated number of people killed in traffic accidents . We also want to develop time series models for predictions. In our analysis, we have used ARIMA and transfer function modeling. ARIMA models explain a future observation using previous values in the same series. The transfer function model is used when the explanatory variable has an impact on the response variable for values in several time periods and when the residuals are dependent. Our data showed a clear seasonal effect and was not stationary which is a requirement for ARIMA modeling. After an adjustment of our data, it was shown that all estimated ARIMA models based on monthly data resulted in moving average parameters. A future observation is explained in a linear composition from former errors. The transfer function model that was adjusted explains the response variable number of people killed with explanatory variable traffic kilometrage. It could not be proven that former values had an effect on the response variable. The residuals for the model could be explained by an ARIMA-process with a moving average parameter. The transfer function model is the model that we found most appropriate and we recommend it for future development. The estimated predictions from our models show no indications that "Nollvisionen" will be fulfilled without radical changes in traffic. The number of people killed will, according to the predictions, stay at the same level as the recent years.