General systems for analyzing instructional interaction have found the most common teacher behavior to be asking questions. This evaluation compares and contrasts two systems for analyzing teacher questions: Price-Belland, developed by the authors from the Bloom-Saunders tradition, and Hough-Duncan, modified for detailed question analysis. Comparisons were made 1) on the nature of the decision-making process required in coding, 2) on the kind of information derived from each system, and 3) on the interpretability of data displays derived from each system. In comparing data obtained from using the two observation systems simultaneously, significant differences were found between systems in the percentage of activity assigned to five of seven common categories. It is recommended that work be continued on developing a more reliable questioning-behavior analysis system. (RT)