The ultimate goal for vehicle aerodynamicists is to develop vehicles that perform well on the road under real-world conditions. One of the most important metrics to evaluate vehicle performance is the drag coefficient. However, vehicle development today is performed mostly under controlled settings using wind tunnels and computational fluid dynamics (CFD) with artificially uniform upstream conditions, neglecting real-world effects due to road turbulence from wind and other vehicles. Thus, the drag coefficients computed with these methods might not be representative of the real performance of the car on the road. This might ultimately lead engineers to develop design solutions and aerodynamic devices which, while performing well in idealized conditions, do not perform well on the road. For this reason, it is important to assess the vehicle’s drag as seen in real-world environments. An effort in this direction is represented by using the wind-averaged drag. The wind-averaged drag coefficient is obtained by weighting the vehicle’s drag under different yaw conditions. However, being computed in idealized environments with low level of turbulence (both wind-tunnel and CFD) it neglects the natural variability of the wind experienced on the road. High yaw angles are only possible in strong crosswind and are therefore more sensitive to natural wind variability. Therefore, it is fair to say that the current wind-averaged drag relies on unphysical test conditions (high yaw angle and low level of turbulence) to assess the vehicle’s performance.In this paper, unsteady aerodynamics simulations based on the Lattice Boltzmann Method (LBM) are used to assess the aerodynamic performance of a detailed SUV vehicle. Using a surface response method, the drag coefficient of the vehicle is fully and efficiently characterized for different yaw angles and turbulence intensities of the environment. This information is then fed into a driving-cycle calculation and show that the increase in the drag coefficient with yaw and turbulence intensity leads to a reduction of the vehicle’s fuel efficiency. A simple loss model is proposed which enables the computation of the energy consumption in real world by pragmatically taking into account the variation of the drag with wind speed, yaw and turbulence intensity. This provides a way to evaluate vehicle designs early in the development cycle and to identify those which are likely to perform well across a wide range of on-road conditions and not just in idealized conditions.