This study investigated how instructional goals, activities, and methods for grading in college level courses vary across disciplines and how these course design variables are related to student ratings, class size, and course level. The study used a 44-item, multiple choice questionnaire mailed to 1280 instructors teaching 2700 course sections in the spring and fall quarters of 1991. Surveys for 887 course sections taught by 486 instructors were returned. Student ratings of instruction for the 887 courses taught by the survey respondents were part of a routine course evaluation process. Patterns of significant association among survey variables appeared which were consistent with disciplinary differences. For example, courses in the engineering-math-science area tended to emphasize fact and concept learning goals, and lectures, and based a high percentage of student grades on exams. Courses in humanitites, on the other hand, emphasized skills development, papers, group discussion, and practice quizzes, and relied less on exams for grading. A similar pattern involving the same course design variables was associated with overall ratings of amount learned, and overall teacher effectiveness. Generally, courses with higher student participation and feedback were associated with higher ratings. A heavy reliance on midterms and finals, and low-frequency feedback grading methods, were associated with lower ratings across disciplines. (Author/JB)