Based on the collected wheat yields in the years of 2008-2013 in the Guanzhong Plain, China and the drought monitoring results of vegetation temperature condition index(VTCI), the Morlet, Mexican Hat and Paul(m
=4) were used to study droughts. Wavelet power spectra of the three non-orthogonal wavelet functions were applied to analyze the multi-time scale characteristics and the cross-correlation degrees of the wheat yields and the VTCIs at the main growth stages of winter wheat. Linear regression models between the yields and the weighted VTCIs at the main growth stages were compared for selecting a better wavelet function for assessing drought impact. The results show that the oscillation energy of the VTCIs using the same wavelet function is different. There are differences of the main oscillation periods determined by three wavelet functions at the same growth stage of wheat, and further there are differences in the wavelet cross-correlation coefficients. The time series VTCIs at the four growth stages of wheat all have a 6-year main oscillation period. The Paul(m
=4) wavelet is most applicable to analyze the multi-scale correlation between the VTCIs and wheat yields, and assess the multi-scale drought impact.