Intra-refresh in video coding is an efficient error control tool of video encoder in minimizing error propagation caused by unreliable network. However, when channel becomes reliable, intra-coded macroblocks (MBs) tend to increase quantization distortion. In this paper, we propose a method of adaptive intra-refresh rate (AIRR) by obtaining the optimum number of intra-coded MBs in a given frame based on the condition of the channel. This is achieved by estimating end-to-end distortion at the encoder side which allows the encoder to adjust the refresh rate before encoding. In this work, we propose a quantization distortion model which estimates source distortion as functions of residual information and quantization parameter (QP). The residual information is estimated using the proposed mean-absolute difference (MAD) prediction model based on the linear relationship between intra-refresh rate and MAD. The proposed models are used to implement a joint source-channel AIRR (JSC-AIRR) scheme using standard H.264/AVC encoder. Accurate estimate of the actual distortion at various refresh rates are achieved and able to estimate the distortion before encoding the frame. The proposed scheme is compared to random intra refresh scheme with various refresh rates and periodic intra refresh scheme. Average improvements of up to 1.54 dB in PSNR video quality are measured and subjective quality improvements are observed which proves the effectiveness of the proposed scheme especially in time varying channel conditions.