The aim of this Thesis is to analyze the driving behavior of food delivery drivers, to identify and evaluate the potential risks they face during their work, and to determine the causes of these risks. To achieve this, a questionnaire was developed, and 200 food delivery drivers from all over Attica were asked to respond based on the stated preference method for various hypothetical scenarios involving changes in delivery time, accident risk reduction and profit loss per delivery. In each scenario, drivers were given three alternative choices: a) Drive more carefully, b) Drive a bit more carefully, or c) Make no changes in their driving behavior. Subsequently, a multinomial logistic regression model and a GLM model were used to process the questionnaire data. The results of the multinomial logistic regression model revealed that the variables influencing the food delivery drivers’ choices in hypothetical scenarios include: delivery time, accident risk reduction, profit loss, age of the driver, the number of fines received, and their opinion on stricter penalties as a measure to improve road safety.