Recent critiques of the usefulness of the concept of class (Pahl 1989;
Clark and Lipset 1991) have developed into more specific criticisms o
f the lack of theory underlying 'class analysis' as practised by Goldt
horpe and his associates (Pahl 1993; Rose 1993). In this paper we argu
e that an adequate theory of class needs to show why classes, defined
in the particular way they are, have some claim to be a basis of socia
l power of sufficient importance to justify the emphasis placed on the
m by class analysis. To answer this question requires that two things
be demonstrated: first, that there is an association between class and
a range of outcomes; and, second, that the mechanisms giving rise to
such associations be specified. Class analysis has little difficulty i
n answering the first requirement but very great difficulties in satis
factorily addressing the second. In discussing these problems we seek
to suggest fruitful future directions for the project of class analysi
s.