Although many programs exist for computing tetrachoric correlations, these programs typically lack one or more of the following features: (1) efficient processing of a large number of variables and computing of a matrix of coefficients; (2) a matrix-smoothing capability; (3) an accurate, reliable computing algorithm; (4) the ability to handle missing data; and (5) nominal cost. TETCORR, which combines an algorithm developed by Brown (1977) with starting values obtained from Divgi (1979a) for more efficient computation, was created to address these and other user concerns.