The hypothesis that memory support (MS) reduces state anxiety (A-State) and errors in a computer assisted learning task was investigated. It was predicted that high A-State students given MS would make fewer errors than high A-State students given no memory support (NMS). Low A-State students were expected to perform equally as well with or without MS. Sixty male undergraduates were randomly assigned to MS or NMS conditions. All students received the same learning materials. The MS group was allowed to see their previous incorrect responses to each problem before attempting it again, whereas this information was not available to the NMS group. There were no significant effects of memory support on A-State as measured by the State-Trait Anxiety Inventory. However, the predicted memory support x A-State interaction was found for errors. These findings were compared with the results of previous research on anxiety and memory support. (Author)