0ne of the most significant developments in writing research over the last fifteen years has been the large number of naturalistic studies of college-level writing in the disciplines in-spired by the WAC movement. In this chapter, I selectively re-view some of the over one hundred studies to suggest what conclusions we might tentatively draw at this stage and what avenues for further research they open. Qualitative studies have predominated in recent years because the early attempts to perform quantitative experimental studies yielded confusing results (for reviews and analyses, see Schumacher and Nash; Ackerman; Geis ler). When these studies attempted to test a central claim of WAC, that writing improves learning or thinking (Emig), they found that writing does not automatically improve either. Indeed, when writing was used to improve students' performance on the usual kinds of school tests, it often had no effect or a negative effect. Ackerman concludes his meta-analysis of twenty-six studies thus:" Writing simply may not perform well in the relatively brief and unrelated learning episodes that appear both in research and in practice"(359). When students were given tasks differing significantly from" the standard knowledge-transmission purposes of the schools," however, writing helped students learn (Geis ler 48). Simple factbased learning may be better achieved through other study strategies (Durst; Penrose). But when students need to learn to solve what psychologists call" ill-structured problems," where there are no single right answersas in most professional workplaces-