Background: A new short-term risk assessment instrument, the Mechanical Restraint–Confounders, Risk, Alliance Score (MR–CRAS) checklist, including three subscales with altogether 18 items, has been developed in close collaboration with forensic mental health nurses, psychiatrists’ etc., and shows evidence of being comprehensible, relevant, comprehensive and easy to use for assessing the patient’s readiness to be released from mechanical restraint. Aim: The aim of this study was to investigate whether the subscales: confounders, risk and parameters of alliance constituted separate subscales and needed further revisions. Materials and methods: MR–CRAS was field-study tested among nurses, nurse assistants and social and health care assistants in 13 Danish closed forensic mental health inpatient units, and a Mokken analysis of scalability and a Spearman correlation analysis were performed. Results: MR–CRAS was completed by clinicians in 143 episodes of mechanical restraint, representing 88 patients, with a mean duration of 63.25 hours. Most patients were younger men, diagnosed within the schizophrenia spectrum. One-third of the patients had repeated mechanical restraint episodes ranging between 2 and 8 episodes. MR–CRAS and especially the parameters of alliance were perceived as usable for assessment of the patient’s readiness to be released from mechanical restraint. The psychometric analyses showed that the three subscales were unidimensional. Conclusions: The study shows evidence of the construct validity of MR–CRAS among clinicians at closed forensic mental health inpatient units. MR–CRAS contributes with a common language and structured, systematic and transparent observations and assessments on an hour by hour basis during mechanical restraint.