The distributed optical fiber Raman temperature sensor system (DTS) uses optical fiber Raman scattering effect to measure temperature. However, the spontaneous Raman scattering signals are particularly weak, completely submerge in the noise, and affect the system’s temperature measurement accuracy. To solve this problem, this paper proposes to use wavelet transform, which has been widely used in engineering field, for signal processing. By using Matlab and VC combined programming, wavelet denoising is applied to the system actually. The results show that this method can eliminate signal noise effectively. The average error of the measured temperature is reduced from 3.1 ℃ to 1.1 ℃. And then, both the resolution and precision of the measured temperature of the system are improved.