Tests to assess problem-solving ability being provided for the Air Force are described, and some details on the development and validation of these computer-administered diagnostic achievement tests are discussed. Three measurement approaches were employed: (1) sequential problem solving; (2) context-free assessment of fundamental skills and knowledge; and (3) constrained tasks. Computer administration was chosen for its advantages of: (1) dynamic control; (2) variable response mode; (3) capturing responses and scoring; and (4) test security. Disadvantages were considered to be outweighed by these aspects. The Diagnostic Achievement Tests consist of Part I, a computerized sequential problems test, and Part II, the Enabling Skills Test (an independent assessment of each skill). Subtests are discussed briefly, and the 15-step process of test development is summarized. Special problems in the development of these specific tests are discussed, including excessive test length, glitches in the computer software, and flaws in the test orientation and approach to obtaining airmen's hypotheses about the locations of faults as they work through the problems. Since initial pilot testing was informative and supportive, validity studies will be done. Four figures support the process description. (SLD)