The unexpected discovery that certain chemotherapeutic agents used in the treatment of childhood cancers have neurocognitive side effects has prompted a search for techniques that identify those medications that place children at risk. An animal model for the assessment of resultant neurocognitive toxicity is described which makes use of simple classical conditioning. We have shown that rats learn about environmental events more slowly following neonatal administration of methotrexate. The changes after methotrexate exposure are not related to stimulus characteristics or to perceptual abilities, but rather to damage to the neural systems involved in acquisition, retention, or recall. Similar problems with learning have been observed in children treated with methotrexate. An effective animal model such as the one described here may help detect and avoid antineoplastic agents that produce severe cognitive defects in childhood cancer patients.