This paper summarizes an investigation of applications and issues in free response (FR) testing during 1989. It draws on ideas from the results of the National Educational Longitudinal Study 1988 (NELS:88) field test, a seminar series at the Educational Testing Service (ETS), working papers prepared for several FR testing applications, and conversations with several individuals at the ETS and elsewhere who have been implementing FR testing. Focus is on incorporating FR items in the NELS:88 test battery. Benefits of the FR format, scoring considerations, resource costs, domain coverage, reliability considerations, validity considerations, equating issues, and implications for longitudinal testing are considered. The primary advantage of the FR format is that it offers the opportunity to assess what the student can do, rather than only what the student knows. However, given the short testing time available for the NELS:88 survey, the use of FR items would necessarily entail a very sparse sampling of the content domains being tested. The resulting low reliability of the change score measurements would limit the usefulness of the data in identifying factors related to educational achievement. It is concluded that FR tests may be more accurate than multiple-choice tests, but they involve considerable cost and sacrifice reliability. The appropriate choice or mix of FR and multiple-choice testing modes depends on the purposes and uses of the test scores. (RLC)