Racism can affect health by making people ill, exacerbating existing illness, and by inequality in access to and utilisation of health services. Recent British publications assessing the methods used in studies of ethnicity primarily considered the status of the variables 'ethnicity' and 'race' and advised on the use of appropriate categories. Such scrutiny of ethnicity research is welcomed, yet authors rarely emphasise the importance of racism as a variable. This paper discusses why racism matters as a variable and poses suggestions for its absence from British health services research. Reference is made to US research to demonstrate that this focus is important and feasible. Health services research that considers ethnicity and excludes the effect of racism may result, at best, in an incomplete understanding. At worst, this omission could itself be perceived as a racist practice.