OBJECTIVES: Health care costs are rarely directly observed for clinical trial subjects. What can be observed is medical resource use. Designing a resource costing strategy involves determining the degree of detailed resource data to collect and identifying unit prices for those specific resources. A higher level of detail can create a more specific estimate of costs, but this strategy increases investigator burden. Our objective is to lay out the conceptual framework of resource costing, draw implications, and provide recommendations for weighing the tradeoffs of the design decisions of resource costing. METHODS: For simplicity, assume that medical costs are composed of two types of resources, X and Y. Unit prices for resources X and Y are denoted px and py. Now consider a bundled resource unit, Z where Z = X+Y. The unit price for Z is pz = px + w*py where w = proportion of X in price population. From this notation, we determine differences in incremental costs estimates depending on whether these estimates are based on resource use of bundled unit Z or its detailed elements (i.e., X and Y). RESULTS: The bundling strategy will produce a different estimate of relative incremental costs between treatment groups if the proportions of each element of the bundled resource item are not equivalent between treatment groups. Absolute incremental costs will differ if the mix of resources in the sample is different than the mix of resources in the population from which the price is estimated. CONCLUSIONS: Bundling to reduce the burden of very detailed resource use data and price weights is justified if the resources are bundled into resource units that are not composed of items that are used more intensely in one treatment group. Price adjustments should be considered if the relative resource intensity is different between the price population and the trial population.