Pharmaceutical R&D is time-consuming, extremely costly and involves great uncertainty. Although there is a broad range of literature on statistical issues in clinical trials, there is not much that focuses directly on the modelling of pre-clinical research. This thesis investigates models and associated software for improving decisionmaking in this area, building on earlier work by the same research group. We introduce a class of adaptive policies called forwards induction policies for candidate drug selection, and show that these are optimal, with a straightforward solution algorithm, within a restricted setting, and are usually close to optimal more generally. We also introduce an adaptive probabilities model that allows the incorporation of learning from a project’s progress into the planning process. Real options analysis in the evaluation of project value is discussed. Specifically, we consider the option value of investing in clinical trials once a candidate drug emerges from pre-clinical research. Simulation algorithms are developed to investigate the probability distributions of the total reward, total cost, profitability index and the required future resource allocations of a pharmaceutical project under a given allocation plan. The ability to simulate outcome distributionsmeans that we can also compare the riskiness of different projects and portfolios of projects.