The laterality index (right - left)/(right + left) has been widely used to measure cerebral asymmetry. However, its validity rests on the assumption that confounding effects of overall head and body size are symmetrical on both sides of the brain. We have found that this assumption may not always be justified. In magnetic resonance imaging data from a group of 43 right-handed normal control and schizophrenic subjects, the laterality index controlled poorly for the asymmetrical confounding effect of height, leading to a spurious (although statistically significant) correlation between height and apparent cerebral asymmetry. Regression modeling is advocated as a more rational and effective way to explore and control size effects confounding cerebral lateralization.