Background: Simulation training can prepare trainees for clinical practice in laparoscopic surgery. Training on box trainers allows for simulation training at home, which studies have shown to be a feasible method of training. However, little research has been conducted into how to make it a more efficient method of training. Our aim was to investigate how box trainers are used in take-home training to help guide the design of take-home training courses. Methods: This study was designed using a mixed methods approach. Junior doctors participating in a laparoscopy curriculum, which included practising at home on box trainers, were invited. Quantitative data on training patterns was collected from logbooks. Qualitative data on the use of box trainers was retrieved from focus groups and individual interviews. Results: From logbooks, we found that 14 out of 18 junior doctors mixed their training modalities, and four practised first on box trainers then on virtual reality simulators. Twelve practised only at home, while five practised at both places and one practised solely at the simulation centre. After a delayed start, most practised for some time, then had a period without training and then started training again towards the end of the course. We found that the themes of the interviews were: training method, training pattern, feedback and self-regulation. Participants identified the lack of feedback as challenging but described how self-rating provided direction during unsupervised training. Mandatory training elements affected when and how much participants practised. Conclusions: When participants practised at home, they took an individualised approach to training. They mixed their training at home with training at the simulation centre. Participants practised at the beginning and towards the end of the course. Self-rating helped to guide unsupervised training where feedback was not accessible. Curricular requirements and testing determined when and how much participants practised.