This article describes a photovoltaic-battery microgrid system used for isolated sites. Indeed, a 50 kW photovoltaic panel is associated with a boost converter. To guarantee more reliable and economical energy supply, a battery storage system is included within the microgrid system. To determine the optimal sizing of the microgrid system, many approaches have been proposed. However, these approaches need a big database applied throughout a year, which can make the sizing more expensive and required a long computing time. In this article, a new approach is proposed, using real weather data (solar radiation and ambient temperature). It is based on two technical criteria; the deficiency of power supply probability and the relative excess power generated to study the system reliability and two economic criteria; the energy cost and the total net present cost to study the system cost. Compared to other approaches, sizing results applied for a year can be deduced using just the month when the microgrid system can work at the optimum conditions with optimal configuration (photovoltaic power and battery) achieving a minimal cost energy without deficiency of power supply probability, which make the sizing simpler and less expensive. The results obtained prove the effectiveness of the proposed approach in sizing problem. All results have been checked carrying out MATLAB simulations using real weather data.