OBJECTIVES: Indirect comparisons are often based on binary outcomes (e.g. relapse / remission) or continuous outcomes. In these cases logistic or linear models are applied to make the indirect comparisons. However, sometimes datasets contain multinomial outcomes, such as ‘complete’, ‘partial’ and ‘no’ response in oncology, that need to be indirectly compared. With multinomial data, different indirect comparisons methods may be required to answer different research questions. Our goal was to identify and qualitatively compare the different techniques that have been used to model multinomial data in an indirect comparison framework. METHODS: A systematic review of the PubMed database was conducted to identify different methods for handling multinomial data in a meta-analysis. Key words included ‘meta-analysis’, ‘ordinal’, ‘ordered’, ‘multinomial’ and ‘proportional odds’, in various combinations. Models were qualitatively compared according to their assumptions, flexibility and complexity. RESULTS: The systematic review identified three methods: a proportional odds model, an ordered logistic model, and a multinomial model. The proportional odds model has a natural interpretation of the treatment effect, is flexible in terms of handling data with different numbers of categories, but relies on the proportional odds assumption. The ordered logistic model also has a natural interpretation of the treatment effect, but increases in complexity when handling data with a large number of categories. The multinomial model’s interpretation for the treatment effect is difficult, but it can model a large number of categories and can handle unordered competing risks and time dependent data. CONCLUSIONS: There are three methods for incorporating multinomial data in a meta-analysis framework with various advantages and disadvantages. Selection of the appropriate model appears to be most dependent on the characteristics of the dataset. We determine that there is sufficient cause for future research focusing on a quantitative comparison of these different methods.