The purpose of this paper is to evaluate the formulation and diffusion of the Japanese style of diet, one of the most important problems in Japanese food policy. It is said that the Japanese style of diet is healthy, but the long-term effects have not been evaluated. Using statistical criterion, I report on this subject. The data used in this paper is Japanese nutritional intake, which I use to formulate a time series model. I chose the ideal PFC balance to measure the formulation and diffusion of the Japanese style of diet. It is useful to show the mutual relationship of proteins, fats and carbohydrates, so I use a vector error correction model. I also consider cointegration to show that long-term relationship. As a result of the estimation, the hypothesis of stability was rejected when I used the data from 1955 to 1995. So I divided the data and tested again. The results of the estimation show that the ideal PFC balance has diffused from 1975 to 1995.