The Texas post-Furman death penalty statute restricts capital punishment to a limited category of murders. If the defendant is found guilty of one of these crimes, the jury must address two and sometimes three questions in the punishment phase of the trial. Affirmative answers to the questions by all jurors result in an automatic death sentence. A “no” answer to any question results in an automatic life sentence. One of the three questions is whether the defendant presents a continuing violent threat to society. From 1974 to 1988, niney-two capital murderers had their sentences commuted to life imprisonment. These commutations allow a “natural experiment” to assess the predictions made by jurors that these individuals would present a future violent threat to society. Patterns of institutional and post-release behavior of this group were compared to similar patterns for defendants convicted of capital murder who were not predicted to be dangerous and who received life imprisonment over the same fifteen-year period. We found that although most capital offenders were model inmates, two commuted capital prisoners committed second murders, one while in prison and the other while in the community. We conclude with a discussion of the validity of current death statutes that require jurors to predict future dangerousness.