Abstract: A real-time LWD data processing method was proposed to address the "black spots" and "pull straight" problems caused by nonuniform sampling of real-time LWD data. Firstly, singular points of real-time LWD data were detected and eliminated utilizing the thin-layer threshold method together with peak-peak or valley-valley ratio method,and the "glitch" phenomenon caused by high-frequency oscillation of LWD was eliminated to improve the signal-to noise-ratio of real-time LWD data.Secondly,the gridding analysis was performed to the real-time LWD data that was obtained.The data gridding and equal interval resampling were accomplished through the establishment of the resampling model of real-time LWD data.The proposed method was applied in the actual LWD data processing.The comparison of processed real-time LWD data with the memory data showed that the maximum relative error and average relative error were 10.92 percent and 8.73 percent respectively,6.06 percent and 3.46 percent less than conventional method;the correlation coefficient was 0.980 8,0.040 2 more than conventional method.The results showed that the combined method of the thin-layer threshold method and the peak-peak or valley-valley ratio method together with Akima method could successfully address the "black spots" and "pull straight" problems.