Relapsed schizophrenic men randomly assigned to a milk- and cereal-free diet on admission to a locked ward were released to an open ward considerably more rapidly than those assigned a high-cereal diet (p = 0.009). When gluten was secretly added to the cereal-free diet this difference did not occur. Release of non-schizophrenic patients was not related to diet. These findings support the hypothesis that ingestion of cereals may be pathogenic for those with the geno-type for schizophrenia. However, results such as these obviously require rigorous independent testing.