Paul Willis's theory of cultural production, first elaborated in Learning to Labour and defended in later work, suffers from two errors, essentialism and dualism. These are more serious than the objections against which Willis has tried to defend his theory-that it is left functionalist and politically pessimistic--and indeed are the reason why he cannot answer such objections satisfactorily. The consequence, notwithstanding Willis's disavowals, is an entrenched romanticising of both 'resistance' and 'culture', especially male working class culture. Perhaps the chief characteristic of the version of cultural studies developed by the Centre for Contemporary Cultural Studies has been an emphasis on the distinctness of the 'cultural level' and a tendency to look to culture--especially working class culture-as the major source of developments toward progressive social change. In the area of education and youth policy the work of Paul Willis has, of course, been provocative and influential. It also exemplifies the particular CCCS concern with and, I suggest, romanticising of, 'culture'. In this article I want to examine Willis's theory in its own right, rather than CCCS work generally. Willis has been concerned to rebut criticisms of Learning to Labour (LL) which allege that it presents as functionalist and therefore as politically pessimistic an account of schooling and working class culture as the accounts of reproduction theorists who preceded and have been criticised by him. In his rebuttals (1981, 1983) he reiterates LL's claims to have introduced a dimension of agency, contestation, and resistance into discussions of reproduction. In his defence he points to cultural creativity and resistance to disown functionalism. On the other hand, he has stated that any 'simple class cultures', in this case youth cultures, should not be romanticised, and acknowledged that 'resistance' may make matters worse, rather than better or merely, as LL seems to suggest, leave (or reproduce) them as they are. In more recent work, dealing with the explosion of youth unemployment since the period documented in LL, Willis has continued to argue