In student-centered online courses, it is important for instructors to monitor, track and analyze behavioral and activity patterns of students. Data of student online activities in an introductory soils course at Iowa State University is continually recorded in server logs. The data were used to identify differences in student learning styles, strategies and patterns of learning in the virtual online environment and to study any relationships between selected variables and student performance in the course.  Clustering techniques were used to classify students based on common shared characteristics. The following variables were used to describe and classify the characteristics of students in the various clusters: login frequency, average time spent on weekly units, frequency of accessing review materials, day of the week quiz is taken, number of quiz repeats, weekly quiz scores and final exam scores.  Predictive models were used to identify the most important variables which influence student performance in the course.