Electrocardiogram　(ECG)　indicators　are　widely　used　in　driver　mental　workload　studies.　Because　ECG　is　always　recorded　continuously　in　experiments,　ECG　indicator　data　often　have　trend.　To　make　follow-up　data　analysis　faster　and　more　reliable,　ECG　indicator　data　should　be　preprocessed　to　remove　or　extract　the　trend　according　to　the　aim　of　the　experiment.　However,　most　researchers　tend　to　ignore　this　point.　Based　on　time　series　theories,　this　paper　proposed　a　set　of　methods　to　preprocess　ECG　indicator　data　series.　In　this　paper,　the　data　preprocessing　method　was　described　in　detail.　And　an　example　was　given　to　illustrate　and　prove　the　validity　of　the　method.