Identifying with more than one ethnicity is becoming increasingly common in New Zealand. It is therefore increasingly important that the methods of outputting ethnicity used by government agencies and researchers effectively incorporate complex ethnic identities. The aim of this research was to assess how important the method of outputting ethnicity is to understanding outcomes in New Zealand, using tobacco smoking as an illustration. Prioritised, total response and single/combination methods were compared. The research used the 2013 Census data for the usually-resident population over the age of 15. Both descriptive analysis and multinomial logistic regression analysis were used. Prioritised and total response outputting generally produced similar estimates and patterns for smoking status. Single/combination outputting indicated that combination ethnic groups tend to have smoking outcomes between those of the constituent groups. Outputting method had a considerable impact on the size of some ethnic groups. Prioritised outputting is inappropriate for use by government agencies and for policy-relevant research principally concerned with children, young people, Māori or Pacific Peoples. Single/combination outputting should be used cautiously but is a useful tool for understanding the associations between ethnicity and outcomes.