Drinking water mains are often categorized by water utilities into two categories, reticulation water main (RWM – diameter small than 300mm) and critical water main (CWM – diameter equal to or larger than 300mm, or at critical locations). The former constitutes the majority of water main assets both in terms of amount and length. Accordingly, most of the water main failures occur on RWMs, causing significant social and economic cost. In this project, we (Data61 & Sydney Water) aim at a statistical failure rate prediction model that can accurately forecast the expected number of RWM failures for a specified future time interval, based on RWMs’ historical failure records, intrinsic characteristics and environmental factors. The model will be the foundation of an optimal maintenance strategy that can dramatically improve the maintenance efficiency of water main networks.