The primary challenge of model-based structural health monitoring (SHM) of long-span cable supported bridges is the development of efficient numerical models that can accurately predict vibration characteristics which are sensitive to uncertain model parameters and structural modelling assumptions. Various model updating techniques with different computational demands and complexities are available in the literature. However, using ambient vibration data, which are often recorded over long periods and characterised by levels of noise and randomness, presents unique challenges to the updating process. To address these challenges, this chapter presents the model updating of a 380m-main span cable-stayed bridge using the real-time SHM data from the structure. Firstly, SHM applications to bridges and model updating methods are reviewed, followed by presenting a typical model updating procedure based on weighted least-squares optimisation and parameter sensitivity approaches which is used in this chapter. While this procedure is well known for model updating applications, it has yet to be applied when using SHM data from a long-span cable stayed bridge where monitoring data variability is inherent. As such, the weighting matrix can be directly defined using this data variability without the need for estimation, which is more effective. While not wholly stochastic, this sensitivity-based updating method maintains the right balance by compromising between the uncertainty and variability in the dataset and parameters in an intuitive way, while avoiding the computational burden of using more complex methods, thus making it more effective and accessible to SHM practitioners. Next, the case study bridge and the acquired SHM data are presented along with the original finite element model with initial modal property results. Finally, the model updating results are discussed demonstrating that the updated model is in good agreement with the modal properties identified from the monitoring data.