A genetic-algorithm (GA) procedure has been ad hoc implemented to obtain a tool for both design and characterization of rare-earth-doped optical amplifiers and lasers. In particular, the routines performing the selection, crossover, mutation, and elitism operations have been written with the aim to investigate the optimal erbium-doped amplifier or laser configuration. Conversely, the GA can be employed in device characterization to identify those parameters of the erbium-energy-level transitions, which are not directly measurable, e.g., the cross-relaxation and up-conversion coefficients. The GA appears intriguing because of its efficiency and versatility. Its operation strategy is noticeably for the capability to identify solutions in complex multidimensional spaces. In this paper, the GA application for modeling and characterizing erbium-doped photonic-crystal-fiber amplifiers is described in detail.