Loring et al. (Journal of clinical and experimental neuropsychology, 2005:27;610–617) observed relationships between VSVT hard item performance and IQ and memory indices in epilepsy surgical candidates, with a potential confound of low FSIQ on VSVT performance. The present study replicated the Loring et al. study in a larger sample and extended their findings by examining the relationships among VSVT performance, FSIQ, and working memory. A total of 404 patients with medically intractable epilepsy completed a comprehensive neuropsychological assessment. Differences in WAIS-III and WMS-III performance were examined as a function of VSVT hard score categories as determined by Grote et al. (2000)--that is, valid, > 20/24; questionable, 18–20; or invalid, < 18. Quantile regression models were constructed to compare the strength of the relationship between FSIQ and VSVT at various points of the FSIQ distribution. Linear regression analyses examined working memory as a potential mediator between FSIQ and VSVT performance. The invalid group performed more poorly than the valid and questionable groups on multiple measures of intelligence and memory. The strength of the relationship between FSIQ and VSVT hard item performance decreased as FSIQ increased, and working memory mediated this relationship. Results suggest VSVT hard item scores may be impacted by working memory difficulties and/or low intellectual functioning.