Introduction: A rigorous mathematical analysis of asset condition data sets, specifically, quantifying their robustness, making the best use of the information contained in them, and setting recommendations for sampling strategy. This will inform both the continued use and ongoing development of investment modelling tools such as health indices for improved network management and design.

Objectives: Specifically the research solves the problem of the best way to estimate the health index frequencies for a whole population of assets, when a sample of only some of the assets has been inspected. The delivered methodology, the “regression for proportions approach” has the flexibility to cope with a number of situations, for example where the inspections have occurred over a number of years. It gives good results and is straightforward to implement.