Computational Fluid Dynamics (CFD) methods are acknowledged as a reliable tool for the assessment of a ship’s calm water resistance. However, CFD methods are known for their very high computational requirements. During the preliminary ship design stage, resistance estimations are traditionally performed using systematic series, empirical methods and formulae, or when possible, using potential flow methods to reduce computational cost, leaving detailed analysis with higher accuracy methods for a later stage. The objective of the present paper is to present an innovative approach for the calculation of calm water resistance, combining Computational Fluid Dynamics (CFD) simulations and Artificial Neural Networks (ANNs). ANNs are considered as a special type of machine learning and the proposed paper aims to demonstrate their potential to replace computationally expensive CFD methods with a much faster and yet sufficiently accurate method to be used during preliminary design stage, when the assessment of large numbers of design alternatives is required. To this end, a large number of hullforms have been elaborated on the basis of a design-of-experiment (DoE), and CFD simulations provided the training data for the construction of the ANN and for the testing and validation studies for the evaluation of its accuracy and the generalization. The development and training procedure of the ANNs, along with typical application results from the evaluation of distinctive ship types will be presented and discussed.