Intelligent tutoring systems (ITS) aim to improve upon previous computer-aided instruction programs by embedding much of the expert knowledge that good human teachers possess. While ITS research has a practical goal of developing programs that assist students in learning, there is little data on their instructional effectiveness. This paper presents preliminary student outcomes from the initial introduction of a prototype intelligent computer-based algebra tutor in a California high school. The focus is on student characteristics, including their background and attitudes about computers and computer-based instruction, and their evaluation of the tutor's usefulness for learning. Pre- and post-semester algebra achievement tests and final course grades provide data on student learning for 80 participants. It was concluded that: (1) positive attitudes toward learning with computers, experiences with this algebra tutor and previous computer experience appear to be unrelated to actual achievement in algebra; (2) there is a significant negative relationship between programming and overall algebra achievement; (3) gender differences existed; and (4) the preliminary data tell little about the relationships between student characteristics, overall achievement, and the nuances of learning algebra with an intelligent tutoring system. (Appended is the student questionnaire and achievement test.) (PK)