Background: There is little data to inform the treatment of severe obsessive compulsive disorder (OCD) in an inpatient or residential setting. Aims: This paper aimed to: a) describe treatment outcomes at a residential unit over 11 years; b) investigate whether treatment was successful for a subset of severe treatment refractory residents; c) compare an intensive treatment programme to a “standard” treatment programme; and d) find predictors of self or early discharge from the unit. Method: We compared treatment outcomes for (i) a minimum 12-week treatment (hereafter “standard”) programme versus a 2-week intensive programme and (ii) for severe treatment refractory cases on the standard programme. We identified 472 residents with OCD admitted to the Anxiety Disorders Residential Unit at the Bethlem Royal Hospital between 2001 and 2012. Outcomes were measured with the Yale-Brown Obsessive-Compulsive Scale (Y-BOCS), Obsessive Compulsive Inventory (OCI), Beck Anxiety Inventory (BAI) and Beck Depression Inventory (BDI) obtained throughout treatment and up to one year after discharge. Results: Although residents had very severe OCD on admission, sequential assessment with the Y-BOCS, OCI, BAI and BDI demonstrated that scores on all outcome measures significantly decreased from pre to posttreatment and were generally maintained at follow-up. There was no significant difference between those on the standard or the 2-week intensive programme. Sixty-nine per cent of residents with OCD made significant improvements, with at least a 25% reduction on the Y-BOCS. There were predictors of self or early discharge but none for outcome on the Y-BOCS. Conclusions: The data support the principle of stepped care for severe OCD.