This work addresses the challenge of modeling tire behavior for virtual vehicle development, particularly in untripped rollover scenarios where a vehicle rolls over solely due to the tire-road friction interface. Despite the existence of various models, limited attention has been given to tire testing methods that accurately represent rollover conditions. We propose a revised lateral tire testing method that uses data from rollover-critical driving tests to derive conditions for a flat-track test bench and that focuses on operational and thermal conditions. These data are used to parameterize empirical Magic Formula Tire 6.2 models, emphasizing the importance of accurate measurements for model parameterization. Component-level validation through comparison of measured and simulated tire reaction forces and moments demonstrates improved estimates of lateral forces, hence lateral friction coefficient, and overturning moments compared with models based on outdoor tire test data with limited operating conditions. At the full vehicle level, the proposed method significantly reduces the error in rollover key performance indicators based on wheel lift-off during Fishhook maneuvers. These findings are particularly relevant for battery electric sport utility vehicles with increased vehicle mass and wheel loads.