PurposeConcerns have been raised about mental illness (MI)-linked disparities in restrictive housing (RH) placements, but research has not compared disparities across multiple types of RH. Differences by MI status in prison adjustment or repeated misconduct could create larger disparities for long-term and highly restrictive types of RH; conversely, more stringent procedures governing their use could lead to smaller disparities for those types. MethodsUsing data on six types of RH in a large southern prison system and logistic multivariate multilevel models, we compared the associations of MI with multiple RH types. Mediation models examined whether infractions, violence, and victimization accounted for any associations. ResultsDisparities were present for all examined types of RH. The largest proportional disparities were found for the rarer long-term solitary confinement placements. Although disparities were smaller for short-term types of RH, those affected more people. The mediators appeared to be better explanations of short-term rather than long-term RH placements among persons with MI. ConclusionsRH placement disparities by MI are largest for types that are of long duration and involve severe restrictions on privileges and activities. In addition, the disparities in those RH types may be less accounted for by commonly examined behavioral factors.