Estimation of longitudinal load such as road slope and rolling resistance is essential for precise cruise control of automotive vehicle. In this paper, using model based road slope estimation method with gravity sensor, precise longitudinal dynamic model of automotive vehicle is presented for auto cruise control. The recursive least square method is implemented for parameter estimation and real-time bias adaptation of g-sensor. Continuously Variable Transmissions (CVT) can cover a wide range of ratios. This makes it possible to operate a combustion engine in more efficient working points than stepped transmission. With this merit, fuel optimal cruise control by CVT ratio control is possible using precise longitudinal dynamic model using model based road slope estimation. In this paper, for fuel optimal auto-cruise control of CVT, the CVT ratio controller is designed for spark ignited engine to tracked best optimal operation line and for vehicle to maintain constant vehicle speed. For two control objectives and nonlinear longitudinal powertrain model of CVT, sliding mode control method is used. The stability and performance of controller are verified by numerical analysis and simulation result.