The United States adopted DRG based prospective payment system (PPS) in order to control the inflation of health care costs. No study used statistical test while many studies reported the cost containing effect of the PPS. To study impacts of the PPS on the Medicare expenditure, this study set the following three hypotheses (1) The PPS decelerated the increase in the hospital expenditure (Part A), (2) the PPS accelerated the increase in the expenditure of outpatients and physicians (Part B), (3) the increase in total expenditure was decelerated inspite of the spill over (substitution) effect because saving in the Part A expenditure were greater than losses in the Part B expenditure. The dependent variables are per capita hospital expenditure, per capita Part B expenditure, and per capita total expenditure for the Medicare beneficiaries. An intervention analysis, which added intervention effect to the time series variation on the Box-Jenkins model, was used. The observations included 120 months from 1978 to 1987. The results are as follows : (1) The annual increase in the per capita Part A expenditure was $5.11 after the implementation of DRG where as that before the PPS had been $11.1. The effect of the reduction ($5.99) was statistically significient (t=-3.9). (2) The spill over (substitution) effect existed because the annual increase in the per capita Part B expenditure was accelerated by $1.73 (t=1.91) after the implementation of the PPS. (3) The increase in the total Medicare expenditure per capita was reduced by $4.26 (t=-2.19) because the spill over effect was less than cost savings in the Part A expenditure.