Abstract A method is presented for deducing porosity and saturations from well log data, where key reservoir parameters, such as water resistivity, may be missing. The method combines routine petrophysical methods with statistical learning, allowing for well log analysis to be performed on a wide variety of wells and geologies. By solving a combination of fundamental physical laws and empirical models and exploiting the character and relationship of the physical laws, one is able to calculate porosity and saturations from well log data and to identify oil and/or gas producing zones. The methodology allows for fitting model parameters when additional well log data, such as core porosity and saturation, are available. A statistical learning approach allows for the probabilistic derivation of porosity, saturations, and fit parameters, including formation water resistivity and the Archie cementation and saturation exponents. The methods have been applied to a number of oil reservoirs and have undergone both manual and automatic validation. The method has demonstrated the ability to quantify porosity and fluid saturations and to identify oil zones, with results comparing favorably to those from traditional petrophysical methods. Because statistical methods are used to derive model parameter fits, output log curves are presented probabilistically. This allows the end user to better understand the range of porosity and saturation possibilities at the zone, well, reservoir, and field levels. The statistical approach also indirectly takes into account noise in the data.