Autism is a highly heritable condition, but there is strong epidemiological evidence that environmental factors, especially prenatal exposure to immune challenge, contribute to it. This evidence is largely indirect, and experimental testing is necessary to directly examine causal mechanisms. Mouse models reveal that prenatal immune perturbation disrupts postnatal brain maturation with alterations in gene and protein expression, neurotransmitter function, brain structure and behavioral indices reminiscent of, but not specific to, autism. This likely reflects a neurodevelopmental spectrum in which autism and schizophrenia share numerous genetic and environmental risk factors for difficulties in social interaction, communication, emotion processing and executive function. Recent epidemiological studies find that early rather than late pregnancy infection confers the greater risk of schizophrenia. The autism literature is more limited, but exposures in the 2nd half of pregnancy may be important. Mouse models of prenatal immune challenge help dissect these observations and show some common consequences of early and late gestational exposures, as well as distinct ramifications potentially relevant to schizophrenia and autism. Although nonspecificity of immune-stimulated mouse models could be considered a disadvantage, we propose a broadened perspective, exploiting the possibility that advances made investigating a target condition can contribute towards the understanding of related conditions.