In this study, a model using the thermal resistance network method and the recursive input estimation algorithm was developed to analyze two-dimensional inverse heat conduction problems. The recursive input estimation algorithm consisted of the Kalman filter and the real-time least squares. A two layered hollow cylindrical tube was assumed with the convection heat transfer boundary condition at the outer wall of the tube. Using the model developed in this study, an unknown time-varying heat flux at the inner wall of the tube was predicted from the known temperatures at some positions on the outer wall of the tube in the longitudinal direction. In the process of predicting the unknown heat flux, the temperature distribution in the two-layered hollow cylindrical tube wall was also obtained. The present model was able to accurately predict the unknown time-varying heat flux profiles of square waves or sinusoidal-like waves with and without spatial variations in the longitudinal direction. (C) 2016 Elsevier Ltd. All rights reserved.