Absconding from acute psychiatric wards is a significant clinical problem that can place patients and others at risk, as well as being burdensome and anxiety provoking for staff. Previous studies have not convincingly demonstrated the best way to minimize the frequency of absconding. The aim of this trial was to evaluate the impact of an intervention to reduce absconding by patients from partially locked acute psychiatric wards. Five acute psychiatric wards in one hospital were entered into a stepped, before-and-after controlled trial. Following 3 months at baseline, nursing staff on the wards were trained in the intervention and monitored in its execution for the next 3 months. Absconding and violent incidents were recorded by nursing staff through shift reports and validated against officially collected forms. Absconding reduced by 25% overall during the intervention period, a fall which was statistically significant. Three out of the five wards implemented the intervention effectively and two of these achieved decreases in their absconding. The other two wards were not able to consistently implement the intervention, and their absconding rates remained unchanged. The findings support the efficacy of the intervention in reducing absconding. Further research is now required to replicate these findings, and to confirm that any reductions are maintained over time.