Three United States Air Force job inventory procedures were studied in depth, categorizing task statements by duties, interviewing technical advisers for job information, and detecting bias in survey samples. Variations in the usual grouping of supervisory tasks were found to be occasionally necessary. Depending upon the career ladder surveyed, nonsupervisory tasks are best organized by work section, function, or equipment format. Criteria are suggested for determining the point at which interviews become more productive than publication research. Interviewing is more effective when advisers are chosen according to experience, rank, kind of organization to which assigned, and job type. An analysis of trends in 10 survey samples showed few unexplainable underrepresentations. (Author)