A number of different systems of classification of ethnicity in the United Kingdom exist. These schemata have been the subject of debate, including about the nature of ethnicity, how or whether it can be categorised, and the relationship between ethnicity, race, and nationality.
The 1991 UK census was the first to include a question on ethnicity. Field trials had started in 1975 to establish whether a question could be devised that was acceptable to the public and would provide information on race or ethnicity that would be more reliable than questions about an individual's parents' birthplaces. A number of different questions and answer classifications were suggested and tested, culminating in the April 1989 census test. The question used in the later 1991 census was similar to that tested in 1989, and took the same format on the census forms in England, Wales and Scotland. However, the question was not asked in Northern Ireland. The tick-boxes used in 1991 were "White", "Black-Caribbean", "Black-African", "Black-Other (please describe)", "Indian", "Pakistani", "Bangladeshi", "Chinese" and "Any other ethnic group (please describe)".
Sociologist Peter J. Aspinall has categorised what he regards as a number of "persistent problems with salient collective terminology". These problems are ambiguity in respect of the populations that are described by different labels, the invisibility of white minority groups in official classifications, the acceptability of the terms used to those that they describe, and whether the collectivities have any substantive meaning.
A number of academics have pointed out that the ethnicity classification employed in the census and other official statistics in the UK since 1991 involve confusion between the concepts of ethnicity and race. David I. Kertzer and Dominique Arel argue that this is the case in many censuses, and that "the case of Britain is illuminative of the recurring failure to distinguish race from ethnicity". Aspinall notes that sustained academic attention has been focused on "how the censuses measure ethnicity, especially the use of dimensions that many claim have little to do with ethnicity, such as skin colour, race, and nationality".
In 2007, Simpson and Bola Akinwale also studied the stability of individuals' responses to ethnic group questions between the 1991 and 2001 census. They concluded that the membership of the "White" category was stable, whereas 7–9 per cent of those in the "Asian" group and 23 per cent of both the "Caribbean" and the "African" group in 1991 had switched to another group by 2001. They suggested that conscious changes in affiliation explained little of this instability, whereas unreliability of the question was significant, partly due to the ambiguous nature of the categories used and partly due to imprecision in the imputation of missing values.