Four types of lateral preference (hand, foot, eye, ear) were measured in 73 patients, who underwent sodium Amytal speech testing. In 63 cases, speech was on the left side whereas in 10, speech was on the right. All of the patients with right hemisphere lesions had speech represented in the left hemisphere. Only a third of the patients with left hemisphere damage prior to the end of the first year of life continued to have speech on the left side. By contrast, 82% of those with left hemisphere damage after this sensitive period had speech exclusively mediated by the left hemisphere. In short, the occurrence of right hemisphere speech dominance without an early lesion to the left hemisphere is rare. The majority of patients with right-sided motor and sense organ preferences had speech represented in the left hemisphere, independent of location or age of onset of the damage. By contrast, in individuals with a clinical history of early left hemisphere injury, left-sided motor and sense organ preferences were linked to right hemisphere speech dominance. In those cases without evidence of early cerebral damage, most of the non-right-sided individuals had speech in the left hemisphere. However, the proportion of unilateral, left hemisphere speech representation in the right-handed and right-footed was higher than that observed in the left-handed and left-footed, even when those with early onset of damage were excluded from analysis. It was suggested that in the normal population, handedness and footedness are relevant factors in predicting cerebral speech dominance. Knowledge of eye and ear preference may be of little predictive value. Further, in a neurological population the overall congruency across the four patterns of lateral preference may provide useful information regarding cerebral speech dominance. A small group (5) of patients with bilateral representation of speech was also examined. Contrary to the experience at the Montreal Neurological Institute, most of these patients preferred the right limb and sense organ. Differences in investigative approach may account for the conflicting observations.