A dichotic listening test, composed of 30 CVC[consonant-vowel-consonant]-word pairs, was administered to 20 adults with aphasia and 20 normal adults. The subjects with aphasia were selected for inclusion in 1 of 4 experimental groups which differed according to initial severity of aphasia and time post onset. All experimental subjects were given a diagnostic language test within the first 4 wk following aphasia onset and again at the time of dichotic testing. Difference scores provided a quantitative estimate of language improvement. There was a strong left-ear preference (-0.363) for the experimental subjects which differed significantly from a right-ear preference (+0.290) obtained from the normal control subjects. Initial severity of aphasia was a significant variable, capable of influencing the magnitude of ear preference. Multiple regression and partial correlational methods revealed significant, positive relationships between the magnitude of ear preference and initial severity of aphasia and between the former and the magnitude of language improvement. Time post aphasia onset was not a significant variable. The theoretical and potential clinical relevance of the results were discussed.