Primary school children with average intelligence and no oral language comprehension deficits—but who were 18 or more months behind their peers in reading comprehension—served as subjects in an evaluation of two approaches to reading remediation. Half the children received teacher‐based tutoring using the DISTAR programme (Science Research Associates, 1983). The remaining students received practice on four computer games. These games were designed to improve performance on a set of information‐processing components shown in previous research to have an important impact on reading comprehension. Training in both conditions focused mainly on word decoding and phonics. Although almost all students improved their reading comprehension test scores after training, the poorest readers made significantly greater gains in the componential training condition than in the DISTAR condition. These results demonstrate the effectiveness of the componential approach with a sample of specifically reading‐disabled children. They also show that componential training can be a practical adjunct to a traditional tutoring programme.