The development of job-specific tests (JSTs) for two occupations is discussed. A reading comprehension test and a mathematical reasoning test were developed for Customs Inspectors, and a reading comprehension test was developed for Social Security Claims workers. JST items incorporated reading samples or math problems from those found on the job. Each job-specific reading test contained 40 items, and the Customs math test contained 30 items. Panels of subject matter experts rated tasks and test items. Correlational and factor analyses that related the two reading tests and the math test to cognitive or non-cognitive marker tests showed that the JSTs were cognitive tests that measured traditional verbal and mathematical abilities. Studies of the Customs tests with about 4,500 job applicants have confirmed the high reliabilities and generally good validities of the tests. The Claims worker test was not used operationally. Effect sizes for the Black (n=about 1,000) and Hispanic (n=about 1,000) Customs Inspector applicants were all close to one standard deviation with respect to the majority White group (n=about 2,500), which is typical of group differences associated with cognitive ability test scores. Research into alternative means of examining job applicants to reduce group differences indicated that combinations of interviews and tests, one of which should be a general cognitive ability test, can reduce group differences without losing test accuracy. Three tables present study data. (SLD)