OBJECTIVES: To assess existing incentives for indication subgrouping in NICE reimbursement submissions using a simple probabilistic model. The model allows for uncertainty and makes explicit assumptions regarding market size, price, subgrouping, cost-effectiveness, market share and reimbursement acceptance. METHODS: The Office of Fair Trading in the UK has recently recommended reform of the UK Pharmaceutical Price Regulation Scheme from profit limitation to one of value-based pricing. It has been debated whether such a move would encourage the targeting of cost-effective subgroups, or whether further incentives would help the UK National Health Service receive a greater share of net health benefit. We have extended an existing probabilistic model of manufacturer pricing decisions to encompass a continuum of patient cost-effectiveness. Uncertainty can be incorporated at the patient and market level. A revenue probability distribution function was derived as a function of price. With this model, we explored the incentives that already exist for the targeting of cost-effective subpopulations using a probabilistic assessment of risks to product revenue streams. We calculated the revenue probability density function under a range of subgrouping scenarios assuming the drug is priced such that, over the whole indicated population, the ICER is £30,000 per QALY. RESULTS: We produced summary statistics from the probability density functions indicative of the level of risk to the product revenue available to the manufacturer, dependent on a number of assumptions. In an example, targeting 20%, 80% and 100% of an indicated population produced expected annual revenues of £16.0m, £51.7m and £14.1m respectively. Probabilities of generating revenues exceeding £15m were 91%, 80% and 17% respectively. CONCLUSION: The revenue probability distribution function provides a useful tool to guide pharmaceutical pricing decisions, according to a company's risk profile. With rational, probabilistic assessment of risk to product revenue streams, it can be shown that incentives already exist for indication subgrouping.