The impact of wind gustiness on the evolution of wave fields is analyzed by superimposing to the nominal wind speed a fluctuation whose amplitude is related to the local air-sea temperature difference. The use of fluctuations represented by a Gaussian process, characterized by coherence in time, produces realistic time series whose characteristics are compatible with those obtained from previous studies and open sea measurements. For the sake of forcing a wave model two more representations of gustiness have been used, a simple alternate oscillation of fixed amplitude around the mean value and a Gaussian process without coherence in time. The introduction of gustiness leads to an evident average increase of the resulting wave heights, larger in the Atlantic Ocean than in the Mediterranean Sea. The randomness of the wind and, hence, to a more limited extent, of the wave fields implies the possible occurrence of wave heights much larger than expected in a nongusty field. Besides two 6 month hindcasts two ensemble exercises have been done by forcing two storms with 50 different random realizations of the input wind fields. On the basis of the statistics of the outputs the authors suggest for practical applications the use of two runs, without and with noncoherent gustiness. This will provide information on the statistics of the possible significant wave heights. The effect of a variable air density on wave generation has been explored by repeating the hindcasts using air density values estimated from the output of a meteorological model. It is found that in the North Atlantic Ocean this leads to an increase of the wave heights during the winter storms. The effects are more limited in the Mediterranean Sea.