Item response theory (IRT) is a useful and effective tool for item response measurement if used in the proper context. This paper discusses the sets of assumptions under which responses can be modeled while exploring the framework of the IRT models relative to response testing. The one parameter model, or one parameter logistic model, is perhaps the simplest of IRT models. It presumes that only a single item parameter is necessary to represent the item response procedure. This parameter distinction is termed difficulty and given the symbol, beta. All unidimensional IRT models operate on the belief that a single fundamental latent construct (theta) is the chief contributory determinant of the experimental responses to each test's items. In the two-parameter model, the discrimination parameter, or the Greek symbol of alpha, is added. This allows the item characteristic curves (ICCs) for different items to exhibit different slopes. This discrimination parameter allows the modeling of the fact that some items have powerful (or feeble) associations to the fundamental construct being evaluated (theta). The three-parameter model adds one more parameter to the two-parameter model to reveal the reality that the lower asymptote of the ICC in accounting for guessing may well require the acceptance of nonzero values for their effective minimum values. The paper reviews some studies involving the use of IRT and discusses the ways in which IRT benefits research and testing. (Contains 24 references.) (SLD)