The purpose of this thesis topic is to investigate when, why, and how dual check valves (DuCV) deteriorate, and the frequency at which they fail. DuCVs are mechanical valves that permit fluid to flow in one-direction only. Power and Water Corporation (PWC) use DuCVs as backflow prevention devices in the Darwin and Palmerston region to prevent contamination of potable water supply. Illness and death have resulted due to contamination of potable water due to DuCV failure. Further, DuCVs in the Northern Territory (NT) have been recorded to deteriorate more rapidly than DuCVs in southern states of Australia. Disassembly of an active DuCV to determine its integrity is not a viable method as it results in an interruption to the water supply. Therefore, the purpose of this thesis is to establish a non-invasive method to predict failure of the types of DuCVs used by PWC, and investigate the rapid deterioration seen in the NT. The properties of the DuCVs will be ranked according to colour fade, age, volume passed to date, mass by individual component, hardness, flow characteristics, and spring stiffness. Upon inspection of the 204 DuCVs, 59% had failed, with 11% completely missing from the brass housing. It was determined that failure rate increased with unit age. DuCVs installed in 2007, 2008, and 2009 had a failure rate of 68%, 61%, and 46% respectively. Further, there was a direct correlation between failure and the volume of water passed to date through the DuCV, with 0% failure below 2000 kL and 83% failure above 8000 kL. Investigations conducted on the colour, hardness, and spring stiffness of the DuCVs resulted in no conclusive correlation between failure. Analysis of the operating temperature, and chlorine concentration of the DuCVs found evidence of chemical degredation. Once the DuCVs have been ranked, a three-parameter Weibull distribution model was used to predict the mean life to failure, and mean DuCV characteristic passed to failure. The methodology used to catalogue the DuCVs will be a combination of surveys, experiments, and statistical analysis.