Strengths and weaknesses of prevailing evaluation models are analyzed, with attention to the role of feedback in each paradigm. A framework is then presented for analyzing issues faced by evaluators of interactive instructional technologies. The current practice of evaluation relies heavily on 3 models developed over 20 years ago: (1) the objectives-based model of R. W. Tyler (1942); (2) the decision-making approaches exemplified by the models of M. Provus (1971) and D. L. Stufflebeam (1983); and (3) values-based approaches such as those of R. E. Stake (1967, 1982) and M. Scriven (1967, 1972). To varying degrees, these three approaches lack methods for systematizing the feedback of both the evaluated and the evaluators. An interactive evaluation model is suggested that attempts a synthesis of the strengths of the other models informed by the research base of recent developments in cognitive psychology. The model consists of four phases. The first, in which feedback figures strongly, is negotiating a paradigm to focus the evaluation and specify major questions, sources of evidence, and standards. The second is collecting and analyzing data. A third phase is judging alternatives and synthesizing a delivery matrix of recommendations. The final phase is helping the client process decision and choice mechanisms to improve the program. Three figures and one table illustrate the discussion. (Contains 164 references.) (SLD)