The present paper discusses the following problem; what is the best theoretical understanding of the social class distribution of health and mortality? The discussion identifies some theoretical problems. Some of these have to do with the importance of social causation of health on the one hand and health-related social mobility on the other. Each one of these two explanations has its own problems, but they are not mutually exclusive. The class distribution of early death can vary both between countries, between two periods in time and between causes of death. Such variability should be exploited for theoretical reasons. Empirical 'anomalies' should not be dismissed or ignored, but taken seriously. It seems clear that a theoretical over-simplification in analysing class and health will prove to be counter-productive.