Groundwater models are often used to predict the effect of climate change and (over)exploitation of aquifers. In the past 10 years, developments in software and hardware made it possible to create complex 3D models that include density dependent flow and transport of chloride. These models are used to simulate the effects of climate change (i.e. sea level rise) and (over)exploitation of aquifers, with respect to ground water heads and chloride transport. However, these predictions (and especially the speed of changes) are highly dependent on the three-dimensional starting concentration. In many cases this is more important than the hydraulic parameters or boundary conditions, especially when predicting the salinization of drinking water wells. We will show the importance of the initial chloride concentration with a few synthetic and real-world examples. We propose a method to derive a three-dimensional chloride distribution using a combination of different types of data: - (airborne) geophysics - Water samples - Bore logs We show how the coupled groundwater model itself can be used to create a physically plausible starting concentration that matches current and historical measurements, and which combination(s) of measurements are the most effective in reducing the uncertainty of the predictions.