This paper explores the connection between the turbo decoding algorithm and Pearl's belief propagation (BP) algorithm. Turbo codes, introduced by Berrou *et al.* in 1993, are a significant advancement in coding theory. The authors argue that the excellent performance of turbo decoding can be explained by applying Pearl's BP algorithm to the "belief network" of a parallel concatenation of two or more codes. However, the belief network for turbo codes contains loops, which are not handled by Pearl's algorithm, which is only exact for loopless networks. Despite this, the paper demonstrates that BP can be used to derive effective iterative decoding algorithms for various error-control systems, including Gallager's low-density parity-check codes, serially concatenated codes, and product codes. These algorithms are shown to be highly effective, providing a general methodology for designing low-complexity iterative decoding algorithms for hybrid coded systems. The paper also discusses the theoretical and practical aspects of turbo codes, including the role of interleavers and the performance of different component codes.This paper explores the connection between the turbo decoding algorithm and Pearl's belief propagation (BP) algorithm. Turbo codes, introduced by Berrou *et al.* in 1993, are a significant advancement in coding theory. The authors argue that the excellent performance of turbo decoding can be explained by applying Pearl's BP algorithm to the "belief network" of a parallel concatenation of two or more codes. However, the belief network for turbo codes contains loops, which are not handled by Pearl's algorithm, which is only exact for loopless networks. Despite this, the paper demonstrates that BP can be used to derive effective iterative decoding algorithms for various error-control systems, including Gallager's low-density parity-check codes, serially concatenated codes, and product codes. These algorithms are shown to be highly effective, providing a general methodology for designing low-complexity iterative decoding algorithms for hybrid coded systems. The paper also discusses the theoretical and practical aspects of turbo codes, including the role of interleavers and the performance of different component codes.