Speculation that schizophrenia is associated with abnormal brain development, the so-called neurodevelopmental hypothesis, has become so popular that it is rarely challenged in the literature. This paper critically examines the evidence for this hypothesis, taking primarily the "devil's advocate" position. The evidence from neuroimaging studies, from studies of prenatal and perinatal intrauterine events and of premorbid development are circumstantial with respect to brain development, many studies are methodologically flawed, and most do not exclude alternative explanations. Evidence from postmortem studies of anomalous cytoarchitecture in limbic and prefrontal cortices is especially noteworthy, as a developmental defect is virtually certain if artifacts can be excluded. Unfortunately, the studies responsible for these findings have serious methodological limitations. The neurobiological plausibility of the hypothesis, which might have been predicted to be its weakest aspect, has proved surprisingly unshakeable in a recent series of animal studies. Ironically, the principal weakness of the neurodevelopmental hypothesis at present is the clinical database on which it rests.