The literature on structural brain abnormalities in schizophrenia is examined to determine whether these abnormalities represent viable candidate markers of vulnerability to the disorder. A majority of studies agree in finding that schizophrenia patients as a group have significantly larger ventricles and smaller limbic brain structures than normal control subjects, but about 50 percent of patients fall within the range of control subjects on these measures. This result has been interpreted to suggest that structural abnormalities characterize only a subgroup of patients. However, given the substantial degree of normal variability in brain structure between families, the use of biologically unrelated individuals as controls is misleading. Studies that have compared schizophrenia patients with their unaffected first-degree relatives have found a much higher sensitivity rate for ventricular enlargement and reduced limbic volumes (i.e., 70%–100%). This high within-family sensitivity, together with evidence from meta-analytic reviews of a substantial relationship between ventricular enlargement and severity of illness, argues in favor of a continuous distribution of the brain pathology in schizophrenia and against a model in which the pathology characterizes only a subgroup of patients. The structural abnormalities observed in both younger and older patients have been found to be highly correlated with familial risk for schizophrenia and obstetric complications, suggesting that some part of the deviance may be present in the premorbid state and that it may reflect both genetic and environmental etiologic processes. The evidence for specificity of the deficits to schizophrenia is equivocal, but no study has yet compared the within-family sensitivities of morphological measures among the major psychiatric conditions. Additional studies using first-degree relatives and well-defined neuroanatomical measurements are needed to determine which brain regions have the highest sensitivities as indicators of schizophrenia in families.