The review discusses two commonly used polytomous Item Response Theory (IRT) models: (1) the generalized partial credit model; and (2) the graded response model. The research results of comparing an equating procedure using the partial credit model with conventional equipercentile equating procedures are highlighted. The review also briefly describes the PARSCALE computer program, which allows the concurrent calibration of dichotomously and polytomously scored items. A linking procedure that sets the parameter estimates obtained from different dichotomous and polytomous items concurrent calibrations to a common scale is summarized. An application of the polytomous IRT model in a large-scale testing program, the National Assessment of Educational Progress (NAEP), is also discussed.