Scene change detection plays an important role in a number of video applications, including video indexing, extraction of semantic features and, in general, pre- and post-processing operations. This article is about designing and evaluating the performance of a dynamic scene change detector optimized for H.264 / AVC encoded video sequences. The detector relies on a dynamic threshold that adaptively traces different characteristics of the video sequence to increase the accuracy of the entire scheme by correctly locating the actual scene changes. The solution has been proven in proper video sequences that resemble real-world videos thanks to a number of different motion characteristics, and has provided good performance without requiring an increase in the complexity of the decoder. This is a valuable question, considering the possible application of the algorithm proposed in postprocessing operations, such as the concealment of errors for video decoding in typical environments of transmission of error prone video such as wireless networks.