In consumer theory tastes are traditionally treated as given, even though there is a history of economists who think tastes can be ·both the cause and result of economic activities. In this research an attempt was made to identify economic determinants of changes and formation of tastes in the case of food commodities. The broad hypothesis under investigation was that relative prices are an inducing mechanism for taste formation.· Specifically, two hypotheses were investigated: (1) the commodities which have a comparative advantage in production induce formation of relative taste preferences favorable to them; and (2) when the relative availability of commodities changes, as a result of technical developments in production and marketing or by the opening up of international trade, people change their tastes in response to change in relative prices. A critical assumption for this analysis was the existence of a universal preference function which is common for people all over the world and which forms the outer envelope· of country specific taste preference functions. In order to test the first hypothesis a standard demand model modified by adding a taste variable as a demand shifter was used. This model was applied to data for forty-three countries and twenty-two food commodities. The usual variables for this model (consumption as the dependent variable, prices and income· as. the independent variables) are measured as the average for the period 1957-62. The taste variable for each commodity was constructed as a ratio of the production of the commodity to the total food production in a country during 1934-38. This variable supposedly captures the influences of country specific factor endowments and climatic conditions in production, thus reflecting the historical differences in the relative price of the commodity among countries. Econometric results indicated that both the size (the estimate was obtained in an elasticity measure) and the t-value of the coefficients for the taste variable are larger in the case of regressions for individual commodities than when commodities were grouped. Also in the case of commodity groups there was little decline in the fits when the taste variables were omitted from the estimating equations. The inference of these results is that taste preferences across countries are largely similar for broad commodity groups but that there exists considerable differences in the country specify taste preferences in the case of individual commodities depending upon differences in the production patterns of countries. The second hypothesis was investigated through a time series analysis by using poultry versus meats in the case of the United States and rice versus other cereals and fish versus meats in the case of pre- and postwar Japan. For this purpose the standard demand model was modified by adding a taste variable comprised of cumulated sums of the past consumptions of own and substitutable commodities. The logic of this approach was that if changes in tastes are induced by changes in relative prices, it should be possible. to capture the taste changes by changes . in the consumption experience of a commodity relative to that of its substitutable commodities. This is viewed to occur through a process of learning by consumption. In the short-run, consumers respond to changes in relative prices by changing their consumption patterns. As experience with new consumption patterns is prolonged over an extended period, tastes gradually change to adjust to the price changes. Econometric results indicated considerable taste shifts for those. commodities for which the relative prices declined sharply over time, that is, poultry in the United States and fish in prewar Japan. This seems to be a reasonable support for the second hypothesis.