This study investigated the effectiveness of two methods of instruction, Teacher-Based Instruction (TBI) versus Computer-Assisted Language Learning (CALL), in improving undergraduate Arab learners' English reading ability in the three aspects of speed, comprehension and vocabulary knowledge. The Experimental Pre-test/Post-test Treatment Group Design was implemented in both experiments carried out in this study. Two samples of 100 and 150 students for Experiments I & II, respectively, were randomly chosen from two higher education institutions in the Arab world. Each sample was divided into two groups depending on learners' pre-instruction preferences for TBI or CALL methods. After eight weeks of instruction using one method, the students of the two groups exchanged instructional methods for another period of eight weeks. In both cases the learners attended three 90- minute reading lectures per week. Quantitative and qualitative data analysis showed that CALL was significantly more effective than TBI for improving the learners' reading ability in the three aspects targeted. Results showed that CALL was more effective due to different reasons: learners were more motivated to read and they enjoyed reading; CALL made learners' reading progress visible to them through immediate feedback; it fostered learner autonomy and their desire to be in control of the program and it offered the learners a large number of different reading activities to work on. Learners' suggestions for improving CALL were mainly related to increasing the time they can use CALL programs in learning to read. Findings of the study should encourage higher education institutions especially in the Arab world to take considerable steps towards utilising computers in instruction. Even at school level this utilisation should be considered, but further research should be carried out with learners at different age levels and in different regions in the Arab world.