Abstract. Any realistic assessment of research projects must allow for possible failures. Scaling down the Potential gain from each project by its probability of achievement ignores the expenditure on unsuccessful projects and is therefore misleading. The method proposed here is more sound: it uses a concept termed the Survival Probability Function, which is simplytheprobability that a Project will still exist at various times in the future. This function can be derived from estimates for the survival factors and durations of project stages. The function can then be used to calculate the statistics of project successes and failures, and can be employed in a standard Risk Analysis. More simply, if can be used as a weighting function for cash flow patterns. This leads to a comparison method for projects with different risks and different potential returns, and thus to project portfolio selection.