Recent research in the ASPECT (Automotive Seat and Package Evaluation and Comparison Tools) program has led to the development of a new method for automobile driver posture prediction, known as the Cascade Model. The Cascade Model uses a sequential series of regression functions and inverse kinematics to predict automobile occupant posture. This paper presents an alternative method for driver posture prediction using data-guided kinematic optimization. The within-subject conditional distributions of joint angles are used to infer the internal cost functions that guide tradeoffs between joints in adapting to different vehicle configurations. The predictions from the two models are compared to in-vehicle driving postures.