Standard and experimental forms of the Johnson O'Connor Research Foundations Analytical Reasoning test were administered to 1,496 clients of the Foundation (persons seeking information about aptitude for educational and career decisions). The objectives were to develop a new form of the test and to better understand what makes some items more effective than others. Internal-structure analysis of the 26 new items indicated that all but 3 could be regarded as at least adequate on the basis of their correlation with current standard forms of the test. Thirteen items with the highest item-total correlations or the greatest contribution to overall test reliability were selected for the new test version. The relationships of six item characteristics to item quality and item difficulty were investigated, and these suggest that the clarity of the conceptual features among the words or concepts contributes more to item quality than any of the features studied. Suggestions are made for developing a new alternative form of the Analytical Reasoning test, drawing on the items not included on this revised version. Two tables present study data, and three figures illustrate the discussion. (SLD)