As is generally true with multiple analyses of data sets, significance tests of more than one correlation coefficient from a single sample pose a risk of an inflated Type I error. Larzelere and Mulaik (1977) devised a multistage Bonferroni procedure to test the significance of a set of correlations that successfully controls the familywise Type I error rate and that is more powerful than the customary single-stage Bonferroni method. Crosbie (1986) prepared a Pascal program to perform the multistage procedure. Larzelere and Mulaik (1977) and Crosbie (1986) gave clear, concise descriptions of the multistage Bonferroni procedure. In brief, the procedure divides the desired familywise Type I error rate (af) equally among the number of tests to be made at each stage. Thus, in the first stage, each of the m correlation coefficients is tested at the at = a.lm level of significance, at being the per-test Type I error rate. The usual Bonferroni procedure terminates at that point, but the multistage method continues if at least one coefficient has been found to be significant. In the next stage, mk coefficients are tested at the at = atf(m-k) level, where k is the number of coefficients earlier deemed significant. The process continues until no additional coefficients are found significant at a given stage. The Appendix provides a listing of MANYCORR, a BASIC program that performs the multistage Bonferroni procedure with a set of correlation coefficients. Output from the program is highly similar to that of Crosbie's (1986) program. MANYCORR does, however, possess some advantages: it is relatively short, it employs a very accurate method of approximating critical values of the Pearsonian correlation coefficient (r), and it yields two supplementary statistics that Larzelere and Mulaik (1977) recommended for use in research reports. MANYCORR performs two-tailed tests, considering the size but not the sign of the coefficients. It determines whether a ~oeffi cient is significantly different from zero by companng the coefficient with the critical value of r at the relevant at level with v = (n 2) degrees of freedom, where n is the size of the sample.