The paper examines the nature and remediation of reversal errors in five severely learning disabled boys (9-11 years old). Results of four phases (baseline and incentive conditions) are analyzed for timed and untimed performance of letter recognition. Among results cited are that for all Ss, errors in naming occurred with the letters p, d, b, and q, and not for the other letters. Data on the four letters are presented separately for two experiments. Results from both timed and untimed experiments are said to provide persuasive evidence that, in general, the reversal errors made by the Ss were easily manipulated. Findings are explained to support an educational rather than neurological explanation for reversals. Persistence of reversal problems is traced to assumptions by educators that the errors are neurological in origin and that direct intervention is unwarranted. (CL)