A theory of the origins of hand preferences (Annett, 1972) is outlined and its implications for the distribution of cerebral speech in handedness groups are examined. Five studies of dysphasia in right and left handers with unilateral cerebral lesions are compared and found surprisingly consistent. Data from three series found alike in all essential respects are combined to give a substantial number of cases with which to test the expectations of the theory. The agreement is close enough to justify a reconsideration of the model in the light of the evidence and also an analysis of possible reasons for the main differences between series. In spite of the many uncertainties involved in inferences from dysphasia to cerebral speech laterality, it is concluded that the evidence is strong enough to offer a useful test of hypotheses as to the distribution of cerebral speech in handedness groups. Other questions bearing on handedness and cerebral speech are briefly considered in order to show the explanatory range and power of the theory.