The purpose of this research was the creation of a methodology for the application of a Bayesian Belief Network (BBN) to the field of real-time water quality assessment in water distribution systems. False positive and false negative detections of water quality events represent an impediment to the application of real-time water quality assessment tools in distribution systems. A BBN was constructed to analyse experimentally obtained contamination data. The BBN model uses the historical behaviour of surrogate parameters (pH, conductivity and turbidity) to discriminate between natural and contamination states in real-time data. Contaminant events were successfully identified by the BBN. The structure of the BBN allowed the consideration of multiple sources of water quality information concurrently to compute a probability of contamination for water quality time series. The model demonstrates a viable means of addressing the problem of false positive and false negatives in real-time water quality.