In this article, the application of fuzzy adaptive Kalman filter for estimation of well-log data problems is presented The adaptive technics has been an approach to avoid the divergence problem in Kalman filter when the detailed knowledge about a system models are not completed Natural gamma-ray tool (NGT) is a natural gamma ray sonde, which detects the emitted gamma-ray during the formation of thorium, uranium and potassium at different energy levels. A computer is used to measure the concentrations of Th (Thorium), U (Uranium) and K (potassium) in different depths. Regular techniques have some uncertainties in the estimation of reservoir evaluation, regarding to the perturbation. In this paper, it is shown that there may be an adaptive Kalman filter with the fuzzy inference system and Rauch-Tung-Striebel smoother, based on a model with less information, has a better estimation of the concentrations of Th, U and K in vertical variations.