In the present paper, several types of collected data were employed to analyse the causes of turbines shutdown in a grid-connected wind farm. Although the average availability of the considered wind farm exceeds 96%, the individual availability of some turbines does not exceed 92%. In this context, the present paper introduces a novel approach of understanding the turbine standstill and availability calculation. This approach is based on a variation of monthly energy production to weight the shutdown time including the maintenance and fault hours. The calm hours in summer are 60% less than the average calm time for the considered wind farm. The distribution of inoperative hours reveals a 300% difference between the original and weighed times of downtime. On the other hand, weighed times are used to assess the impact of various faults causing turbines shutdown. The frequency distribution of the faults has shown that 42% of turbine shutdowns are caused by network disturbances, 70% of them are attributed to gild disconnections. Finally, the time distribution of the network faults is investigated to illustrate their impact on the turbine standstill. (c) 2005 Elsevier Ltd. All rights reserved.