As healthcare becomes information intensive technology increasingly plays an important role in managing patient care; information gathering and dissemination; and co-ordinating work. One approach to ensuring resources, staff and systems are allocated and used efficiently is process modelling. This paper presents some findings from the Dependability Interdisciplinary Research Collaboration (DIRC) project, on ‘process modelling’ as an aspect of managerial activity and the work involved in both creating and implementing process maps or models. Our ethnographic investigations document some of the problems in developing process maps of medical activity as part of a desire to ‘improve knife to skin time’. We suggest that process maps are not systematic, rational, scientific deductions of the most efficient process. Process maps are locally sensible versions of best practice, and problems may arise where such locally sensible versions are exported throughout an organization to other settings where other relevances may apply.