An approach that is currently gaining popularity in educational measurement is one that treats item response theory (IRT) as a special case of nonlinear factor analysis (NLFA). A brief overview is provided of some of the research that has examined the relationship between IRT and NLFA. Three NLFA models are outlined, emphasizing their major strengths and weaknesses. These are: (1) McDonald's polynomial approximation to a normal ogive model; (2) the factor analytic model for dichotomous variables of Christoffersson and Muthen (1975, 1984); and the full-information factor analytic model of Bock and Aitkin (1981) and Bock, Gibbons, and Muraki (1988). Although the full-information factor analytic model appears to be the strongest of the approaches described, it appears that a greater number of empirical studies should be undertaken to compare the various NLFA models with respect to how accurately they can recover simulated parameter values. An appendix presents an IRT-NLFA proof. (Contains 61 references.) (SLD)