Biased responding on the Sternberg Recognition Memory Test was observed in four patients with traumatic brain injury. None of these individuals met the Diagnostic and Statistical Manual's (DSM-IV) criteria for malingering. Individual recognition memory scores were high shortly after injury, declined to chance or below at the 6- and 12-month evaluations, and then showed substantial recovery by the 24-month evaluation. Recall memory performance actually declined slightly across this same 2-year period. Recognition memory scores were related to the extent to which the patients endorsed somatic items on the Hamilton Rating Scale for Depression (HAM-D). Poor performance was associated with high somatic scores. The relationship between memory and somatic scores on the HAM-D in this case series suggests that unconscious processes can influence memory performance and, because of this, that clinicians should not use such performance as a primary indicator of malingering. More importantly, biased responding and actual memory deficits may coexist. This is indicated in the current cases by the failure of recall memory to improve during the 2 years these patients were followed.