Prediction by simulation of the minimum-time lap of a flat and level racing circuit by a high-performance motorcycle is treated. A novel method is described. Constituents of the method comprise: (i) a high-fidelity mathematical model of the vehicle; (ii) a rider model with control of throttle/brake position, steering torque, and upper-body lean torque. The rider model uses linear quadratic optimal preview control with adaptation to variations in running conditions by gain scheduling; (iii) a circuit model; and (iv) a learning process through which the rider arrives at the best speed target. The constituents are discussed in turn. Then, systematic reduction of lap times for each of two circuits is demonstrated. Aspects of the performance of the motorcycle in very fast laps of the circuits are shown, providing evidence of the effectiveness of the method established.