Currently, it is a common practice within transport companies to reduce fuel consumption of their fleet by awarding best drivers with public symbolic recognitions, as part of their programs to promote efficient driving. Usually fuel economy (FE) is the key performance indicator used to evaluate them. However, FE depends on several other parameters such as the working route, vehicle weight and vehicle technology. Therefore, companies with a diverse fleet composition require a fair KPI to select their best drivers. In this work, we present a model to assess drivers' performance based on FE measurements. Based on multivariate statistical analysis of one-year FE data of an urban and interurban bus transit company, we found that drivers' FE exhibit a normal distribution when they are grouped within three categories: (a) the route, representing the driving cycle; (b) the vehicle age, representing the engine technology, and (c) the number of axles, representing the weight of the vehicle. Thus, the standard statistical analysis to identify outliers was used to identify best drivers and vehicles that require maintenance.