Confession evidence presented at trial is extremely damaging to the defense. This study examines the impact of a recanted confession on jurors' perceptions of a murder case in which the defendant claimed to have falsely confessed due either to an underlying medical condition, a psychological disorder, or the general stress of the interrogation. Also included were an inadmissible confession condition and a no-confession control condition. Results showed that the impact of the confession was mediated in part by the circumstances surrounding it. Although probability-of-commission estimates were as high in all of the conditions involving a confession, conviction rates were marginally higher when the disputed confession involved mental illness or interrogation-induced stress than when there was no confession, yet did not differ when the disputed confession involved a medical disorder or there was no confession. These findings show that not all recanted confessions are treated equally and that people selectively discount confessions depending on biases and beliefs they hold.