The debate over class dealignment is in part a debate about the meaning of class, which cannot be separated from the issue of the relationship between class and voting. Neither the simplified two-class model that has often been used nor the more sophisticated Goldthorpe class schema are adequate either at the conceptual or the empirical level. Both fail to deal coherently with the intermediate positions in the class structure. The argument that `social class' is of continuing significance for the analysis of voting behaviour and of party identification is correct, but only if the nature of the stratification order is properly understood. The Cambridge Scale, a measure of general hierarchical, material and social advantage, based on patterns of social interaction, is shown to be comparable to the Goldthorpe schema in terms of statistical prediction. It is argued that it is preferable in the sense that it most clearly captures the single most important aspect of ‘class’, which is hierarchical position.