This paper concerns an analysis on a set of data based on GDP of Bangladesh during the period 1979-80 to 2007-08 collected from Bangladesh Bureau of Statistic (BBS). Data relating to production and expenditure accounts have been reflected both in current and constant prices. In the source mentioned above, GDP is measured at constant price following SNA-93. We use data which are measured at constant price (base year 1995-96). We try to find a suitable ARIMA model for forecasting GDP and seven sectors of GDP out of fifteen. From our study we find that all the models appear to be satisfactory because the forecasting MAPEs (Mean Absolute Percentage Error) are all very low. But before this one must verify the validation of the model in different time period, because a forecasting model may lose its validity and suitability as time changes.