The present Study introduced a magnitude of error method as a new malingering detection strategy. Thirty university students were requested to either simulate a memory impairment or to try their best. Magnitude of response error was also investigated in a group of 30 nonlitigating moderate to severe closed-head-injured patients, and 7 suspected clinical malingerers. Probability of selection values were calculated across multiple choice items from tests of recognition memory adapted for the WMS-R Logical and Visual Reproduction Memory subtests. These value scores were employed to identify analog and clinical malingerers. Analog and suspected clinical malingerers were much more likely to select low probability multiple-choice items, and often endorsed choices that neither controls nor moderate to severe CHI patients endorsed. Probability of selection value scores were able to classify 86% of the analog and 100% of the suspected clinical malingerers with 100% specificity for controls.