The large fluctuations observed in the daily solar radiation profiles affect highly the reliability of the PV system sizing. Increasing the reliability of the PV system requires higher installed peak power (P-m) and larger battery storage capacity (C-L). This leads to increased costs, and makes PV technology less competitive. This research paper presents a new stochastic simulation model for stand-alone PV systems, developed to determine the minimum installed P-m, and C-L for the PV system to be energy independent. The stochastic simulation model developed, makes use of knowledge acquired from an in-depth statistical analysis of the solar radiation data for the site, and simulates the energy delivered, the excess energy burnt, the load profiles and the state of charge of the battery system for the month the sizing is applied, and the PV system performance for the entire year. The simulation model provides the user with values for the autonomy factor d, simulating PV performance in order to determine the minimum P-m and C-L depending on the requirements of the application, i.e. operation with critical or non-critical loads. The model makes use of NASA's Surface meteorology and Solar Energy database for the years 1990-2004 for various cities in Europe with a different climate. The results obtained with this new methodology indicate a substantial reduction in installed peak power and battery capacity, both for critical and non-critical operation, when compared to conventional approaches applied in PV sizing. (C) 2011 Elsevier Ltd. All rights reserved.