We propose an automated method for inference of porosity and saturation, which can be used for quickly assessing reservoir rock quality and fluid volumes. The method is based on the inversion of multiple well logs using the Bayesian Theory. The well logs consist of neutron porosity, density, sonic (P-wave and S-wave velocities) and resistivity, which are related to porosity and saturation through the appropriate rock physics models. Following the Bayesian Theory, the formal solution of the inverse problem is a posterior distribution, which takes into account data uncertainties and prior information, including clay content, matrix mineralogy and fluid properties. Several tests, using single logs or a combination of logs, show that integration of multiple well logs operates constructively to produce reliable estimates of both porosity and saturation, considering that appropriate resolution is not available from single data sets. Tests include both synthetic and real data, allowing for comparisons with true values and estimates from the conventional method of interpretation. Even with simplistic uncertainty models (i.e., additive normally distributed noise), our Bayesian approach closely matches porosity and saturation obtained by the conventional method.