A study was conducted to compare the writing performance of tenth grade students across three modes of discourse. Subjects were 80 students who had been classified as "masters" or "nonmasters" of writing based on their scores on the writing section of the Texas Assessment of Basic Skills. The students were assigned three writing tasks, each of which elicited a different type of writing--expressive, explanatory, or persuasive. The results of holistic evaluation revealed that student writing performance did vary across modes of discourse, with expressive writing samples receiving the highest ratings, followed closely by persuasive writing samples. No difference was found between the performance of masters and nonmasters. The findings point out the inadequacy of using a single writing sample to judge writing competency and suggest that such items as student interest in the topic, the specific demands of the writing mode, and the variation of criteria embedded in holistic rating guides can lead to fluctuations in student scores. (Holistic scoring guides and student instructions are appended.) (FL)