A continuing concern of theorists in composition and rhetoric has been the problem of how to design effective writing assignments. With the recent movement in writing assessment from objective tests to writing samples, this problem has become even more important for those who produce "tests" designed to elicit writing samples. A study therefore investigated whether the manner in which an assignment is stated affects the writing outcome. Four assignments (essentially the same assignment stated differently or more elaborately) were given randomly to 127 high school juniors and seniors in six different classes in two midwestern, suburban high schools. The four assignments were designed to test the claims of theorists such as Joseph Williams, E. M. Hoffman and J. P. Schifsky. The first assignment specified an imaginary audience and purpose and offered suggestions for planning and writing steps; the second assignment specified a "real" audience and purpose by "putting something real at stake"; the third assignment specified a general audience and purpose and brief criteria; the fourth did not specify any audience or purpose. Results showed no significant difference in the performance of students; those with the first assignment did as well and as poorly as those given the fourth. A multiple-choice questionnaire administered to the students after the writing test showed they did not value one assignment more than another. (Two tables of data are included; 18 references and the questionnaire are attached.) (TB)