In order to determine the reliability of international students' English as a Second Language (ESL) class performance as a predictor of academic performance, a population of 169 international students at Portland State University (PSU) in Portland, Oregon were selected and statistical tests were performed on their GPAs and TOEFL scores. Individual students' GPAs were computed for all ESL classes taken and for each component--grammar, reading, writing, and speaking/listening--as well as for the end of the first quarter, first year, and second year of academic study. Pearson Coefficient Correlations were then computed for the ESL and academic GPAs. The students were also divided into subgroups based upon gender, nature of academic major (more-verbal or less-verbal), age, nationality (Asian or Middle-Eastern), number of ESL classes taken, amount of previous English-speaking college experience, prior education level, TOEFL score, and PSU entry date. Then mean GPAs were calculated for each of these which were compared by T-tests. The results were mixed. While it was clear that ESL grades and academic grades correlated strongly for some variable groups, it was difficult to determine which variables had the strongest effect because of subgroup composition. For example, female students, students from Asia, and students whose majors fit the more-verbal category showed strong and statistically significant correlations for ESL-second year academic grades; but 90% of the women were from countries of Asia, and the proportion of both Asian and female students in the more-verbal majors was much higher than that of males or students from countries of the Middle East. There were two patterns that stood out in the research results. One was that ESL students who had taken twelve or more ESL classes maintained consistent academic GPAs across time, while those who had taken fewer than twelve ESL classes and all students in the non-ESL group had GPAs that started higher than those of the first group but declined over the two year period--some of them enough to be statistically significant. The second pattern was that groups with strong ESL academic GPA correlations tended to do better in college that did those with weaker ESL-academic GPA relationships. TOEFL scores were correlated to both ESL and academic grades. In the first case, there were both moderately positive and statistically significant relationships. In the second case, the correlations were very low; and for non-ESL students, there was essentially no correlation. But comparing mean academic GPAs showed a significant difference between students who scored below 500 and those who scored 500 and above on the TOEFL.