This paper highlights the need for a systematic, content aware, and theoretically-based approach to test design. The cognitive components approach is endorsed, and is applied to the development of a computerized perceptual speed test. Psychometric literature is reviewed and shows that: every major multi-factor theory includes a clerical/perceptual speed (CPS) factor; and the CPS factor is defined by tests using number, letter, or figural stimuli, but is best identified with several stimulus types. A computerized CPS test was developed, which contains three subtests: random number strings, random letter strings, and nonsense figure strings. Items are presented in a same/different format that minimizes intersubject and intrasubject strategy variation in search sequencing. Latency and errors should increase as the length of the comparison string gets longer. The slope of a simple linear function should increase as the number of elements increases. Therefore, the length of comparison strings was varied from one to seven elements for number and letter strings, and from one to five for figure strings. In order to obtain normative data, the test was first administered to 435 Navy recruits. Following minor changes, the final version was administered to 722 subjects. Each subject took the three computer CPS subtests and three paper and pencil tests. Results indicate that the computer CPS test was a better psychometric instrument than were the paper and pencil counterparts; it controls speed-accuracy tradeoffs, produces greater individual differences, is more reliable, has greater construct validity, is less "g" loaded, and is probably a purer measure of CPS. This paper calls for the resurrection of the cognitive components paradigm, which combines the knowledge and techniques of psychometric, experimental, and cognitive psychology into a mutually beneficial framework for the test design/development. A 38-item list of references, five data tables, and four figures are included. (RLC)