A novel wavelet denoising method is proposed for the pressure surveillance of heavy oil thermal recovery well using white light extrinsic Fabry-Perot interferometric (EFPI) fiber-optic sensor. This method can effectively suppress the non-stationary noises and disturbances in downhole environment. A method for estimating the optimal wavelet decomposition level is proposed based on the analysis of the signal-to-noise ratio (SNR) enhancement varying with decomposition level. No prior knowledge of the real pressure signal spectrum is needed. The optimal wavelet decomposition level is determined by using the downhole pressure data collected in one of the heavy oil thermal recovery wells in Xinjiang, China. Field test results show that the proposed method can increase the SNR of pressure profile by 2.6 dB, and thus further effectively increase the accuracy of the forecasting result of daily oil production.