The aim of this Diploma Thesis is the investigation of the impact of eco-driving on fuel consumption using smartphone data. To achieve this objective, data collected from 15 drivers who participated in a naturalistic driving experiment for a period of six months are analyzed. During the first four months, the participants drove in the way they normally would, followed by two months of eco-driving. The analysis was conducted using the statistical method of lognormal regression. Through the regression models it was examined whether driving characteristics recorded by smartphone sensors affect and can therefore predict fuel consumption. In order to analyze the available data, three statistical regression models forecasting fuel consumption were developed. The first is the overall model, the second refers to the first phase of the experiment (baseline driving) and the third to the second phase of the experiment (eco-driving). The results demonstrated that by improving the participants’ driving style, a remarkable reduction in fuel consumption was observed; and smoother and more ecological driver behavior was achieved. The parameters that affect the fuel consumption were the average acceleration and deceleration, the harsh acceleration events that occurred per kilometer, the harsh breaking events that occurred per trip, the duration of the trip, the average duration of stops and the demographical characteristics.