This paper argues that formative evaluation of instruction, which is generally agreed to be critical for instruction in any medium, is even more crucial when the instruction is to be delivered by interactive technologies such as computers, interactive video, hypermedia, or the various forms of interactive multimedia systems. It begins by discussing formative evaluation as a formal step in instructional development models, noting that the models rarely specify where in the process such evaluation should take place. The foundational assumptions and biases of the paper are then discussed, including the current controversy over qualitative and quantitative research and various issues involved in selecting the research methods to be used. Several types of data collection and analysis methods that can be used to answer important questions concerned with interactive instructional technologies are considered, and the use of a method that is appropriate to answer the particular evaluation questions involved is advocated. A discussion of the benefits of considering alternate methods of formative evaluation introduces a review of the results of evaluations of the overall effectiveness of interactive technology-based instructional programs, primarily computer assisted instruction and interactive video. An overview of planning and conducting formative evaluations as an on-going process through all phases of design and development is then presented. Multiple methods for collecting and analyzing data are also reviewed, with emphasis on the selection of appropriate methods. Suggestions for reporting the results and a summary of some of the major considerations in conducting formative evaluations conclude this paper. (63 references) (BBM)