The aims of this study were to determine faculty's ability to accurately and reliably categorize exam questions using Bloom's Taxonomy, and if modified versions would improve the accuracy and reliability. Faculty experience and affiliation with a health sciences discipline were also considered. Faculty at one university were asked to categorize 30 sample exam questions using either Bloom's Taxonomy or one of two modified versions of Bloom's Taxonomy. Overall accuracy improved when a modified version of Bloom's Taxonomy was used. Collapsing the six categories of Bloom's into three (knowledge; comprehension and application; analysis, synthesis, and evaluation) showed higher levels of accuracy than when each category was collapsed with its neighbor. There was no difference between health science and nonhealth science faculty in accuracy. Overall interrater reliability was low regardless of experience or health science affiliation.