For a thin reservoir, such as the Zechstein Main Dolomite (generally 33–83 m thick) of the BMB oil and gas field of Poland, where the thickness (c. 40 m) is often around a quarter of the dominant wavelength, the composite seismic response results from variations in the petrophysical properties, thickness, lithology, effective pressure and temperature, as well as in the acoustic impedance of the encasing materials. To use the BMB Field 3D seismic data for porosity prediction, 20 post-stack attributes were extracted from a seismic volume, defined by two zero-crossing time horizons that bound the reflections of the Main Dolomite. Because of the large number and the interdependency of the extracted attributes, principal component factor analysis was applied, resulting in the coding of 70% of the variability of the extracted attributes, in six orthogonal factors. Sequential nonlinear regression revealed that the first three factors, F1, F2 and F3, are the significant predictors of porosity. Cross-validation indicated a class of poorly estimated porosities resulting from poor quality/complexities in the seismic data, and a class of good porosity estimates that were subsequently used in a final cross-validation for establishing optimum weights and orders of porosity prediction polynomials. The final cross-validation indicated optimum orders of five, three and two for polynomials in F1, F2 and F3, respectively and optimum weights corresponding to validation well No. 1 (MO-3).