In the spring of 1979, the National Institute of Education, in collaboration with the United States Office of Education, the UCLA Center for the Study of Evaluation (CSE), and members of a nation-wide network of research and development agencies, sponsored a national colloquy on the theme that testing could have an important impact in improving the effectiveness of instruction, but that much remained to be understood about testing needs and problems. Eight regional conferences were held. Each conference involved presentations from national and regional figures in the area of testing and instruction. The conferences also provided an initial training opportunity in test development and test selection to acquaint participants with some of the newer ideas in the field. Each conference devoted its second day to important regional issues and needs. The recommendations of the regional conferences can be synthesized as follows: (1) current testing perspectives need to be refocused; (2) decisions about test development and selection should involve a much broader constituency than is presently the case; (3) the instructional application of testing needs to be refined; and (4) greater coordination is needed among federal, state, and local testing needs. Some implications growing from these recommendations are presented. (Author/BW)