This paper examines audio-recorded data from meetings in which NHS managers decide whether to fund high-cost drugs for individual patients. It investigates the work of a Welsh individual patient commissioning (IPC) panel responsible for sanctioning the purchase of 'un-commissioned' treatments for exceptional cases. The case study presented highlights the changing rationales used for approving or denying a cancer drug, Tarceva, during a period when NICE first suggested it was not cost effective, but then changed its position in a final technology appraisal recommending use when the cost did not exceed that of an alternative product. Our data show how decisions taken in the shadow of NICE guidance remain complex and subject to local discretion. Guidance that takes time to prepare, is released in stages, and relates to particular disease stages, must be interpreted in the context of particular cases. The case-based IPC panel discourse stands in tension with the standardised population-based recommendations in guidance. Panel members, who based their decisions on the central notions of 'efficacy' and 'exceptionality', often struggled to apply NICE information on cost-effectiveness to their deliberations on efficacy (clinical effectiveness). The case study suggests that the complex nature of decision making makes standardization of outcomes very difficult to achieve, so that local professional judgement is likely to remain central to health care rationing at this level.