The concepts of hot and cold are important in disease etiologies and systems of food classification in many parts of the world. A number of writers on hot-cold beliefs have assumed that the classification of foods is the central element in this system of beliefs, and that it is consistent. They have then proceeded to explain these beliefs in symbolic or adaptive terms, generalizing from systems of classification which have only local applicability. More recently a number of writers have recognized the importance of intracultural variation in the hot-cold classification of foods, and have turned their attention to revealing the underlying principles of classification. But because food classifications are only consistent within a limited geographical area, if they are consistent at all, these 'general principles' are only applicable to a single area or limited cultural context. In this article I describe the hot-cold system in a rural area of the Indian state of Gujarat. I show that by proceeding from the classification of diseases, and not from the classification of foods, it becomes possible to reveal certain underlying classificatory principles which also appear to be applicable to other manifestations of the hot-cold system. These principles seem to be based on the phenomena which accompany temperature changes in nature. Finally I suggest that hot-cold beliefs should be seen as an explanatory model which seeks to make the puzzling and threatening phenomena of disease and death more acceptable and predictable.