The purpose of this thesis is to develop a prediction model for fuel consumption by taking into account design, operational and environmental parameters of a typical passenger vessel (Ro/Pax type). More precisely, an ANN predictive model was developed based on 322 historical voyage reports of a typical vessel, elaborating different input variables for the development of the model. After testing 90 ANN models of varying architectures, topologies and combinations of input variables, it was concluded that a Multilayered Feed-Forward neural network model (ML FFNN) with 10-15-1NN structure is the optimal neural network, which can accurately predict the fuel consumption of the reference vessel. The findings also revealed that the model’s highest prediction accuracy was achieved when exogenous factors were used as input variables, indicating that the prediction of fuel consumption is more related to exogenous variables rather than on its previous values, namely autoregressive model. In addition to the above, the performance of the ANN model is compared with a Multiple Regression (MR), and it is observed that the former model seems to have a better forecasting accuracy as its MAPE (2.16%) is lower than the MR’s MAPE (2.54%), denoting also the non-linear relationship between the fuel consumption and the input variables. The proposed FFNN model can be integrated into the energy management system of companies with similar vessels, as it can help ship operators in choosing the most efficient measures in order not only to achieve vessel’s fuel efficiency and sustained operational performance but also to reduce ship-generated emissions, fact that will also lead to lower operational costs for the shipping company. The contribution of the thesis in the literature is the provision of a more accurate method for the prediction of the fuel consumption of this vessel type through the incorporation of several exogenous variables important for the vessel operation.