Abstract: Wireless multipath channels often exhibit block-sparse structure with large delay spread. This paper addresses the problem of estimating block sparse channels in OFDM systems. According to whether the block partition information is available, novel algorithms of orthogonal frequency division multiplexing (OFDM) block-sparse channel estimation based on block sparse bayesian learning (BSBL) framework utilize the sparsity property are proposed. For the purpose of improving the computational speed, the two proposed algorithms use bound optimization (BO) method to learn the unknown parameter which control the sparsity of block-sparse channels. These algorithms based on BSBL framework only use pilots to estimate channel, we also propose joint BSBL algorithms that both the data and the pilot subcarriers are incorporated to improve estimation performance without decreasing spectrally efficiency. Monte Carlo simulations have shown that the proposed algorithms have better performance than the conventional channel estimation algorithms. By using the joint BSBL algorithm, the performance will be further improved.