The EM algorithm is used to derive maximum likelihood estimates for item parameters of the two-parameter logistic item response curves. The observed information matrix is then used to approximate the covariance matrix of these estimates. Simulated data are used to compare the estimated and actual item parameters. (Author/BW)