This paper presents a Lyapunov-based design of repetitive learning controllers for uncertain systems. The controller design is unified due to the use of the parametrization and norm-bounding technique, and the novelty lies in the less requirement for the knowledge about the system undertaken. In addition, this design can handle fractional uncertainties involved in the dynamics effectively. The estimation for the fractional uncertainties is performed to facilitate the controller design and property analysis. Unsaturated- and saturated-learning algorithms are characterized, through rigorous analysis, for the establishment of the boundednesss of the variables in the closed-loop system and the the asymptotical convergence of the tracking error. Numerical examples are provided to verify effectiveness of the proposed learning control scheme.