Maternal immune activation during pregnancy is an environmental risk factor for psychiatric illnesses such as schizophrenia in the offspring. Patients with schizophrenia display an array of cognitive symptoms, including impaired working memory capacity. Rodent models have been developed to understand the relationship between maternal immune activation and the cognitive symptoms of schizophrenia. The present experiment was designed to test whether maternal immune activation with the viral mimetic polyinosinic:polycytidylic acid (polyI:C) during pregnancy affects working memory capacity of the offspring. Pregnant Long Evans rats were treated with either saline or polyI:C (4 mg/kg; i.v.) on gestational day 15. Male offspring of the litters (2–3 months of age) were subsequently trained on a nonmatching-to-sample task with odors. After a criterion was met, the rats were tested on the odor span task, which requires rats to remember an increasing span of different odors to receive food reward. Rats were tested using delays of approximately 40 s during the acquisition of the task. Importantly, polyI:C- and saline-treated offspring did not differ in performance of the nonmatching-to-sample task suggesting that both groups could perform a relatively simple working memory task. In contrast, polyI:C-treated offspring had reduced span capacity in the middle and late phases of odor span task acquisition. After task acquisition, the rats were tested using the 40 s delay and a 10 min delay. Both groups showed a delay-dependent decrease in span, although the polyI:C-treated offspring had significantly lower spans regardless of delay. Our results support the validity of the maternal immune activation model for studying the cognitive symptoms of neurodevelopmental disorders such as schizophrenia.