F. Samejima (1973) proposed a continuous response model in which item response is on a continuous scale rather than some discrete levels. This model has potential because in many psychological and educational assessments, the responses are on a conceptual continuum rather than on some fixed levels. As a first step toward studying the applicability of the continuous response model to psychological and educational assessment, it was important to develop an item parameter estimation algorithm. Because the estimation of item parameters involves an incidental theta parameter, the EM algorithm is used. The EM algorithm is a popular technique for item response theory item parameter estimation that is used in commercial calibration software. The item parameter estimation procedure is described step-by-step. The procedure was programmed in C language, and it was evaluated using real and simulated test data. The real data were from 3 forms of the Work Keys (American College Testing Program) writing test assessment with samples of 7,097, 2,035, and 1,793 people. Simulated data were also based on Work Keys writing test forms for 200, 500, and 2,000 simulated examinees. Results of both studies show that the continuous response model seems applicable to real test data with many categories of responses and that the EM algorithm works reasonably well in recovering true item parameter values. (Contains 2 tables, 3 figures, and 12 references.) (SLD)