Scalable video coding (SVC), an extended version of H.264/AVC, is designed to transmit high-quality video bitstreams over heterogeneous networks. However, such video bitstreams are sensitive to transmission error, thereby severely degrading its quality. Inter-layer prediction in SVC causes errors to be propagated not only to subsequent frames but also to frames in the upper layers, when the errors have occurred in the lower layers. This paper presents an inter-embedding error-resilient scheme to reduce the distortion caused by the loss of inter-layer prediction information in SVC. The proposed algorithm exploits the reversible data embedding scheme to hide essential information of the lower layer without damaging the original data. Experimental results demonstrate that the proposed method provides a better PSNR performance than the frame copy by an average of 4.89 dB in a 2-layer SVC decoder, and an average of 4.54 dB in a 3-layer SVC decoder in the case of whole frame loss.