This paper tackles the problem of sizing a standalone hybrid photovoltaic-batteries-hydrogen (PV-hydrogen) system, by applying an evolutionary algorithm. The system is specifically designed to cover the power necessities of remote, isolated telecommunications facilities, so it must be able to work in an unattended way during at least 2 years. Under this specific constraint, we develop an evolutionary algorithm which optimizes the number of PV panels and their distribution to feed two different arrays of batteries, and also the slope and azimuth of the panels. The well-known simulation program TRNSYS has been used in order to simulate the behavior of the real PV-hydrogen system. The evolutionary algorithm looks for the set of parameters which best performance of the system provide, in terms of hydrogen pressure remaining after two years and cost of the complete PV panels. The performance of the proposed evolutionary algorithm has been tested for the case of a real PV-hydrogen system sited at National Spanish Institute for Aerospace Technology (INTA), Torrejón de Ardoz, Madrid, Spain, where the proposed approach obtained a good solution which fulfils the constraint specifications of the system.