Dissimilar metal welds, where high strength ferritic steel is joined to stainless steel via nickel-based alloy filler weld, represent a challenge for structural integrity assessments of nuclear plants due to the complex microstructure and consequent variation in properties over short length scales. This variation in microstructure and properties gives rise to a non-uniform evolution of plasticity at crack-tips and therefore the measurement of the fracture toughness of dissimilar metal welds is challenging. Local approach to fracture methodologies and more specifically, the Gurson-Tvergaard-Needleman ductile fracture model, provide the ability to simulate ductile fracture in materials, including dissimilar metal welds and can provide insights into the observed fracture behaviour. However, the challenge with using such models is that a number of material specific parameters need to be calibrated. Typically, the parameters are derived from basic microstructural analysis, reference to other published work and/or fitting the model predictions against experimental fracture toughness test data. The current research characterises the microstructural regions in a dissimilar metal weld where SA508 Grade 4N ferritic steel is joined to AISI 316L(N) stainless steel via a filler weld of nickel-based Alloy 82. The associated variation in strength across the key microstructural regions within the dissimilar metal weld have been characterised. A fracture toughness test program was performed by Wood, Warrington using compact tension specimens, in which the notch was positioned on the SA508 Grade 4N to Alloy 82 fusion boundary. 3D X-ray computed tomography was used in conjunction with fractography to characterise the micro-mechanical fracture processes within the tested specimens. The X-ray computed tomography data were then used to quantify the evolution of damage using the variation in void volume fraction beneath the fracture surfaces. These data were used to derive key parameters within the Gurson-Tvergaard-Needleman model. The approach was first used to calibrate the model against reference fracture data, before the calibrated model was used to predict the effect of (a) the position of the initial crack-tip with respect to the fusion boundary and (b) specimen geometry and loading on fracture behaviour, in the absence of residual stress. This research demonstrates the benefits of using quantified 3D void growth data as the basis to derive ductile model parameters, rather than curve fitting, since the predicted accumulation of damage local to the crack-tip is representative of that measured in test specimens. Moreover, the research strengthens guidance relating to the application of local approach models to predict the fracture behaviour of welded structural components.