The results of an evaluation of Vermont's statewide assessment initiative are presented, with information about the implementation of the program, its effects on educational practice, the analytic challenge presented by the portfolio scoring process, the reliability and validity of portfolio scores, and the tensions between assessment and instructional reform. Under development since 1988, the program was first implemented statewide in the 1991-92 school year for mathematics and writing in grades 4 and 8. Information about implementation and impact was gathered from teachers and principals. The experience of the Vermont portfolio program to date suggests the need for patience, moderate expectations, and ongoing formative evaluation. Neither of the basic goals of the program, to use complex performance to measure student performance and to use a performance-assessment program to spur instructional improvement, can be met easily or quickly. The Vermont experience illustrates the tension between these goals and the need to make difficult trade-offs in compromising. Four appendixes discuss reader biases and the generalizability of mathematics and writing data, and they present a scoring sheet. Twenty-six figures and 51 tables present evaluation data. (Contains 19 references.) (SLD)