An integrated multidisciplinary workflow has been implemented for quantitative lithology and fluid predictions from prestack angle gathers and well-log data within the Realgrunnen Subgroup in the Goliat Field, southwestern Barents Sea. We have first performed a qualitative amplitude-variation-with-angle (AVA) attribute analysis to assess the spatial distribution of lithology and fluid anomalies from the seismic data. A simultaneous prestack elastic inversion was then carried out for quantitative estimates of the P-impedance and ratio. Probability density functions, a priori lithology, and fluid class proportions extracted from well-log training data are further applied to the inverted P-impedance and seismic volumes. The AVA qualitative analysis indicates a class IV response for the top of the reservoir, whereas anomalies from the AVA attribute maps agree largely with the clean sand probabilities predicted from the Bayesian facies classification. The largest misclassification in the lithology classification occurs between shaly sands and shales. A mixed lithology and fluid classification indicates a smaller degree of overlap and allows for the discrimination of hydrocarbon sands. Integration of a qualitative AVA analysis and a quantitative Bayesian probability approach helps in constraining the depositional facies variability within the Realgrunnen Subgroup. Finally, a possible influence of tectonic activity during the deposition of the Realgrunnen reservoir is inferred based on the facies distribution maps.