23.5% of high secure patients stayed over 10 years or equivalent. 18.1% of medium secure patients stayed over 5 years or equivalent. Medium secure long-stay prevalence varied wildly from 0% to 50% across units. Gender showed no association with long-stay in either setting. Ethnic class showed no association with long-stay including white vs non-white. Age was higher among long-stayers with one-third over 50. Hospital orders with restrictions doubled long-stay odds versus other MHA sections. Civil sections cut long-stay odds by 42-63%. Prison transfers cut long-stay odds by 68-65%. Admission from prior high or medium secure multiplied long-stay odds eightfold versus prison. Community admissions trended toward long-stay in high secure. Intellectual disability wards had highest long-stay rates versus personality disorder or mental illness. Two-thirds of long-stayers admitted from same or lower security levels. No national standards explain medium secure variation in long-stay. Sociodemographics matter less than service and legal factors.