Trainee medical officers (TMOs) participated in a study comparing three methods of simulation-based training to treat medical emergencies occurring in a hospital setting. The methods were: All groups had the same total teaching time. Participants (n=61) had an initial (pre-training) assessment by written tests, self assessment and simulations of medical emergencies ('VT' and 'HYPOglycaemia'). Participants were tested again post-training using similar simulations to the initial scenario and a new scenario ('ANAphylaxis'). Trained 'experts', blinded to the teaching group of participants, watched video-CDs of the simulations to assess participants' performance. All groups demonstrated increased knowledge and confidence (pre-training scores compared to post-training) but no differences could be detected between the three groups. In simulated emergencies, post-training scores were also improved. There was no difference between groups in the 'HYPO' scenario but in the 'VT' scenario there was moderate evidence that Group 3 was superior. In the 'ANA' scenario, Group 3 had far better test scores, especially in behavioural items. There did not appear to be any significant advantage of using whole body manikins over CSBT and simple part-task trainers. Full-mission simulation training helped develop the ability to recognise when skills learnt to manage one type of medical emergency can be useful in managing another emergency not previously encountered.