Definitions relating to ethnicity and race are muddled and politically contentious. Empirical information is often collected on varying criteria and this makes it difficult for adequate comparisons to be made. We propose to follow the most usual definition of ‘black’ in the British context by using the term to refer to people of Asian, African and West Indian origin. Hence we include people from the Indian sub-continent and their descendants (sometimes excluded in strict definitions of ‘black’), African people and their descendants and, most importantly as far as Britain is concerned, people from the West Indies – sometimes known as ‘Afro-Caribbean’ society or the African ‘diaspora’ (dispersed peoples) – and their descendants. This definition draws a sharp line between ‘black’ people and members of other ethnic minorities (Chinese, Cypriot, Iranian and so on, who may suffer comparable conditions in some respects). Amina Mama, writing in Feminist Review No. 17 on these and other definitional questions (1984: 34), endorses this general approach; it is also the one followed in the recent survey by the Policy Studies Institute from which unique information is available. We shall discuss in due course some of the difficulties created by this definition, but for the moment it seems the most desirable one to adopt. In practice most British statistics break down the category ‘black’ into two groups based on Asian and West Indian origin. In comparison with the situation in the United States the ethnic composition of contemporary Britain falls fairly clearly into the way the Policy Studies Institute (hereafter PSI) entitles its report: Black and White Britain (see Brown, 1984). In the US, the black population is complemented by substantial Hispanic, Pacific Asian and American Indian minorities, rendering the empirical data far more complex. We are not in a position to give an adequate discussion here of either the empirical or the political dimensions of race and gender in the US, although obviously some of the debates in Britain are directly influenced by previous developments there. Nor can we address the situation in the rest of Western Europe, where the patterns of race and ethnicity are different again. The major sources for empirical information are, in the absence of data from the census of 1981 (information about place of birth is not very helpful and the ethnic origin question was in the end rejected because of its political sensitivity), the Labour Force Survey and the recent PSI report. The Government Labour Force Survey, 1983, included a significant set of questions on the basis of ethnic origin as defined by the respondent; it is summarized in Social Trends 15 (Central Statistical Office, 1985). The survey was used to calculate details of the ethnic distribution of the British population (94% white, 1% West Indian, 2% Asian and so on); as far as women are concerned it was of use in that some questions, particularly relating to employment, generated data that was broken down by sex. (The LFS was an improvement on the General Household Survey, which has generated data on ethnicity, but based on the interviewer's assessment of ethnic origin and much less reliable.) The PSI survey has provided completely different, information in that it was an extremely detailed study using very long interviews with a smaller sample (a nationally representative sample of 5,000 black people of Asian and West Indian origin and a nationally representative sample of over 2,000 white people). It is systematically broken down by sex and covers many topics of interest to feminists, such as household composition and so on. In addition to these major sources there are useful figures in Britain's Black Population (The Runnymede Trust and the Radical Statistics Race Group, 1980). In using aggregate figures of these kinds, there is a great danger that the categories that are used (‘white’, ‘Asian’, ‘West Indian’) will be seen as homogeneous. We hope that in challenging old stereotypes we will not be guilty of parading new ones: evidence that one group earns more on average than another does not mean that all its members do, or that they are absolutely well off. Furthermore, aggregate figures such as these may occlude and mask class variation of a more general kind within ethnic minorities. In what follows we attempt to reconsider some of our previous work in the light of the question of race. Although we believe it to be typical of – certainly indebted to – a much wider body of socialist-feminist work, we feel it is easier and less contentious to re-assess our own work than to take on this much wider field. We begin by looking at the relationship of household and wage labour, move on to broader questions of theory and ideology, and conclude with our critique of the family.