AbstractThe Unmanned Space Vehicle test flights will use a 7m 1300kg aircraft. The first three launches will take place at the Italian Space Agency ASI base in Trapani–Milo, Sicily, through a stratospheric balloon that will drop the aircraft at a predefined height. After free fall acceleration to transonic velocities, the parachute deployment will allow a safe splash down in the central Mediterranean Sea. The goal of this article is to show the preliminary analysis results for the first USV flight.We carried out a statistical study for the year 2000–2003, evaluating the typical summer and winter launch windows of the Trapani–Milo base.First, in the center Mediterranean, we define safe recovery areas. They cannot be reached during the balloon ascending phase so, after a sufficiently long floating part able to catch the open sea, the balloon will go down to the release height (24km). The simulation foresees a 400,000m³ balloon and 3 valves for the altitude transfer.A safe splash down must occur far enough from the nearest coast: the minimum distance is considered around 25km. The vehicle should be released at a distance, from the nearest coast, greater than this minimum amount plus the USV model maximum horizontal translation, during its own trajectory from balloon separation to splash down. In this way we define safe release areas for some possible translations.Winter stratospheric winds are less stable. The winter average flight duration is 7h and it is probably too long for the diurnal recovery requirement and its scheduled procedures.Comparing past stratospheric balloons flights and trajectories computed using measured meteorological data (analysis), with their predictions made using forecast models and soundings, we obtain the standard deviation of the trajectory forecast uncertainty at the balloon–aircraft separation. Two cases are taken into account: predictions made 24 and 6h before the launch.Assuming a Gaussian latitudinal uncertainty distribution for the prediction 6h before the launch, we are able to identify the forecast trajectories that have a probability greater than 97% to reach the safe release areas.Simulating the summer windows trajectories for the years from 2000 to 2003 and for the favorable ground wind days, we obtain the number of trajectories with the desired forecast probabilities.