The current study evaluated the utility of the WMS-III Faces I subtest (Faces) for the assessment of malingering. Thirty nonlitigating traumatic brain injury patients and 30 control participants were administered Faces under standard administration and instructed malingering conditions. Although the two groups obtained similar scores when taking the test under standard instructions, both groups produced significantly lower performances when instructed to malinger, indicating that Faces is sensitive to malingering, but less sensitive to traumatic brain injury. The total raw score provided stronger classification accuracy than an empirically weighted combination of the five easiest items (i.e., floor effect items). A raw score cutoff of 31 yielded the maximum classification accuracy with 93.3% sensitivity and 80.0% specificity.