How should one evaluate a program for the gifted? Unique design concerns need to be taken into account when selecting a method for assessing gifted program outcomes. The author reviews design issues and proposes creative solutions to omnipresent dilemmas facing the would-be evaluator.Educational evaluation can take on different forms depending upon the questions asked. For example, sponsors of evaluations of gifted programs might ask one or more of the following questions: &dquo;Who should be served by the program?&dquo;, &dquo;Is the curriculum consistent with the goals and objectives of gifted education?&dquo;, &dquo;Do students become more creative as a result of the program?&dquo;, and &dquo;Is the program worth the cost?&dquo; These questions represent different types of evaluations. Respectively, they are needs assessment, process evaluation, outcome evaluation and evaluation of cost effectiveness.Evaluation design refers to the procedures and conditions employed by the evaluator to collect data. Since data collection is often different across the four major types of evaluations, design issues depend upon the type of evaluation conducted. In this paper, the discussion is limited to outcome evaluations. I have selected this type of evaluation because of the current emphasis on educational outcomes (for example, see &dquo;A Nation at Risk,&dquo; 1983) and because I am an advocate of that approach (Carter and Hamilton, 1985).There are three reasons why evaluation designs must be judged against criteria different from the experimental designs in the laboratory. First, field based evaluations must be designed around administrative constraints such as time, money and logistics. These constraints restrict the evaluator's ability to manipulate and control variables. Consequently, evaluations in the school setting are not as &dquo;tight&dquo; as more traditional laboratory studies. Second, Popham (1975), points out that program evaluations are not concerned with external validity, i.e., the ability to generalize beyond the program setting. Sponsors of program :evaluations desire to know how effective &dquo;their&dquo; programs is in &dquo;their&dquo; setting. Third, it is often impossible to determine whether participant outcomes are due to the program or some other variable(s)-a matter of internal validity (Campbell and Stanley, 1963) -unless controls are imposed which ironically, create a setting that is atypical of the program. In fact, Snow (1974) argues that true program effects can only be determined when a &dquo;representative design&dquo; is employed. Snow recommends designs that are representative of the natural environment because he assumes that experimental manipulation causes the learners to act differently than they would under normal conditions. These reasons illustrate the dilemma of program evaluators. They seek designs that control extraneous variables but do not disturb the natural setting nor violate the school's administrative structure. I have yet to find a design to accomplish this goal. Howev...