The purpose of this study was to determine whether the Criminal Sentiments Scale (CSS), a putative measure of criminal thought content, was capable of predicting recidivism in adult and juvenile offenders. Thirteen nonoverlapping samples from 10 different studies (total N = 1,789) were included in a meta-analysis in which the total CSS scale and its three subscales—Attitudes Toward the Law, Courts, and Police (LCP); Tolerance for Law Violations (TLV); and Identification With Criminal Others (ICO)—were used to predict reoffending. The total score, LCP, and TLV achieved modest to low moderate effect sizes and the ICO achieved a weak, but significant effect size. There were no signs of significant scatter between studies and no evidence of publication bias. It is concluded that the CSS may have value in predicting recidivism, although additional research is required to determine whether it has incremental validity relative to measures of criminal though process.