Based on the premise that composition skills and their evaluation are crucial to the educational process, this paper presents a tentative research program for conducting future English as a second language (ESL) composition evaluation studies. The program developed in the paper covers the following topics as areas which merit further rigorous investigation in ESL composition evaluation: (1) the types of possible reading analyses; (2) the various methods of estimating the reliability of composition ratings; (3) the effects of the different purposes of writing elicited during the testing process; (4) the measurement properties of the rating scales; (5) the estimation of rater effects; (6) the number and the nature of the constructs underlying writing ability; (7) the assessment of errors; (8) the role of writing apprehension in composition assessment; and (9) the need for systematic, replicable analyses of the composing process. After a discussion of the two most commonly used methods of evaluating compositions--holistic scoring and analytical scoring--the paper concludes by noting that although it may be neither feasible nor practical to investigate all of the topics proposed, the results of the proposed studies could augment and enhance an increasing body of ESL composition pedagogy. A list of 35 references and 4 figures conclude the paper. (SKC)