Ethnicity is an important variable in studies of health inequalities in New Zealand. Yet there are ongoing concerns about the nature, quality and use of ethnic data. In 2004, Statistics New Zealand recommended that researchers and policy makers no longer use the system of ethnic prioritisation, a system originally designed to assign people with multiple ethnic responses to one ethnic category. While across a range of disciplines researchers have shifted to using either total ethnic counts or single and combination counts, many health researchers continue to use ethnic prioritisation. Census data show that when using prioritisation there are significant losses to Pacific, Asian and European groups, especially for young people. Losses are especially high for New Zealand born people in all age groups. Health researchers need to consider very carefully the costs and benefits of using prioritised data. Based on the census data we suggest the costs, in terms of loss of information and possible biases in findings, outweigh any benefits.