Spectra of many meteorological data such as wind speed and temperature are time variable. Thus a Fourier analysis is not sufficient. In the present work, the wavelet transform is applied as a time-frequency analysis to the meteorological data for the region of Adrar (27.9 degrees N, 0.3 degrees W. 263 m), Algeria. This analysis is carried out in order to investigate the power spectra behaviors of both temperature and wind speed and their variations with time. To determine the relationship between these two meteorological parameters, the cross wavelet analysis is also applied. The study is carried out using data extending over a period of four years. The analysis is applied over a frequency range from 0.002 to 0.5 cycles per day. The results show that significant synoptic oscillations of periods between 2 and 16 days occur mainly in the cold season in both wind and temperature time series. Those oscillations are characterized by short life durations of one to few weeks. Wavelet power spectrum has also revealed the presence of intra-seasonal oscillations of periods between 30 and 60 days. These intra-seasonal oscillations have been observed mainly in the warm seasons. This study reveals also that temperature and wind speed covary especially at the synoptic and the intra-seasonal frequencies. (C) 2009 Elsevier Ltd. All rights reserved.