It is common that there are more than one solutions when inferring model parameters according to preference information provided by decision makers. As there are some drawbacks when dealing with such situation by traditional ordinal regression methods, robust ordinal regression(ROR) provides another solution taking into consideration all compatible parameters. The applications of ROR methods in ranking and sorting problems are summarized and the necessary and possible preference relations accompanied by their computation methods are analyzed.Then, the combination of ROR and the stochastic multicriteria acceptability analysis is introduced, and the methods for the selection of the most representative model in various circumstances are described. Finally, some conclusions are drawn and the future research directions of ROR in multi-criteria decision making are envisaged.