In this paper, the combined normalized and offset min-sum algorithm is proposed for efficient decoding of low-density parity-check (LDPC) codes in optical and magnetic storage systems. The proposed algorithm can be considered as a general case of the normalized and offset algorithms, and offers better BER performance compared to them. The performance improvement is presented for a partial response magnetic recording channel. Our simulations are based on an LDPC code from the class of finite-geometry LDPC codes. It will be shown that the proposed method has a performance very close to that of the belief propagation algorithm which has a much higher complexity.