A semi-analytical weight estimation method is proposed for composite, oval fuselages, but is also applicable to conventional fuselages and to metallic materials. Loads applied to the fuselage include pressurization, steady-state maneuver loads and inertial loads. The primary structure around the passenger cabin is sized, based on first-ply failure using the Tsai-Wu failure criterion, global and local buckling. Moreover, maximum deflection due to transverse pressure is constrained for skin panels and sandwich panels. Sandwich panels are also sized for crippling and wrinkling. Empirical factors are used to calculate secondary structure and non-structural mass. In order to reduce in-the-loop calculation time, surrogate models of the sizing procedures are used, by means of neural networks. Verification of the sizing optimization was done by comparing the sized members with a genetic algorithm and validation of the failure calculations was done by finite-element analysis. It was found that the proposed method is capable of predicting metal, conventional fuselage mass satisfactorily, with acceptable breakdown of weights and estimated thicknesses. Additionally, the method can be used for unconventional aircraft configurations and composite material. Using composite material, a weight saving of around 15% is observed as compared to aluminum.