Based on the monthly sunshine dataset from the Lishui observation station from 1953 to 2014, the trend analysis method is used to analyze the interannual, decadal and seasonal trends of sunshine duration, and then the Morlet wavelet transform and MTM (MultiTaper Method) are applied to study the periodic characteristics of sunshine duration, and the significant periodic signals are reconstructed by MTM, to further study the lesssignificant periods on different time scales with the oscillation time of evolution. The results show that sunshine duration exhibits very significant decline trend in Lishui, with a rate of 3.5 percent per 10 years. The maximum sunshine duration is in summer, followed by that in autumn. The Morlet wavelet and MTM analysis results show that quasi12year cycle is the most significant in the whole time domain, with regular alternating positive and negative oscillations and the maximum spectral density; in addition, the cycles of 6 years and 2 to 4 years cycle are also apparent. The MTM reconstruction analysis indicates that the amplitudes of quasi12year periodic signals vary with time, and there are three distinct and slowly varying envelopes.