BACKGROUND: Alexithymia is a trait involving difficulty identifying feelings (DIF), difficulty describing feelings (DDF) and externally orientated thinking (EOT). It is a risk factor for criminal behaviour. It is commonly assessed with the Toronto Alexithymia Scale (TAS-20), but the psychometrics of the TAS-20 have not been tested across the range of offender populations, and it has been suggested it might be unsuitable in incarcerated offenders. AIM: To establish the psychometrics of the TAS-20 among incarcerated offenders. METHODS: Factorial validity was examined using confirmatory factor analyses, and the invariance of this factor structure was tested against a published community sample. Reliability coefficients were calculated. RESULTS: One hundred and forty six incarcerated offenders were recruited. The factor structure of the TAS-20 was invariant across the samples. The intended factor structure composed of DIF, DDF and EOT factors performed well overall (with a reverse-scored method factor added), but six EOT items had low factor loadings. The total scale score and DIF and DDF subscales had acceptable reliability, but EOT did not. CONCLUSIONS: Our results suggest that the TAS-20 functions similarly in offender and community samples. Its total scale score, and DIF and DDF subscale scores can be used confidently, but the assessment of externally oriented thinking may not be adequate with this scale alone. In sum, the TAS-20 can facilitate robust assessment of alexithymia in closed criminal justice settings as well as in the wider community.