Meehl's taxometric method has been shown to differentiate between categorical and dimensional data, but there are many ways to implement taxometric procedures. When analyzing the ordered categorical data typically provided by assessment instruments, summing items to form input indicators has been a popular practice for more than 20 years. A Monte Carlo study compared the accuracy of taxometric analyses implemented in the traditional way (without summing items) and taxometric analyses implemented with the summed-input method. These analyses generated no support for the summed-input method, which substantially reduced discriminating power for 2 of the 3 procedures studied. Accuracy was highest when 5 or more indicators and 4 or more ordered categories were used. Findings from the simulation study were then used to help interpret the results for taxometric analyses of antisocial personality disorder criteria with real research data. In this example, the traditional method yielded clearer results than the summed-input method. Implications for the use and further study of the taxometric method in assessment research are discussed. (Contains 4 tables, 5 figures and 2 footnotes.)