In quantitative seismic interpretation we strive to quantify reservoir properties from seismic reflection data. This task is very challenging because we face an underdetermined, non-linear inversion problem with non-unique solutions. Inverse rock physics modelling offers an alternative approach to solving this inversion problem, compared to model-based inversion; e.g. it does not use a background model or wavelet as input to the modelling. It is a probabilistic inversion which is highly based on rock physics, and can combine various rock physics models with different types of input data. It is well equipped to handle the ambiguity of the inversion problem because it scans through the whole solution domain for a given rock physics model. We demonstrate the use of this method on data from a gas discovery in the Norwegian Sea.