We present an analysis of a time series of wind strength measurements recorded at Florence airport in the period October 2002 – March 2003. The data were taken simultaneously by two runway head anemometers, located at a distance of 900 m, at a frequency of 3.3⋅10−3Hz. The data show strong correlations over long time spans of a few tens of hours. We performed an analysis of wind velocity as it is usually done for turbulence laboratory experiments. Wind velocityreturnsand wind velocity differences were considered. The pdfs of these quantities exhibit strong non‐Gaussian fat tails. The distribution of the standard deviations of the fluctuations can be successfully reproduced by a Gamma distribution while the Log‐normal one fails completely. Following Beck and Cohensuperstatisticsapproach, we extract the Tsallis entropic index q from this Gamma distribution. The corresponding q‐exponential curves reproduce with a very good accuracy the pdfs of returns and velocity differences. © 2004 American Institute of Physics