The invention relates to a coding method with incremental redundancy in which it is determined (620) whether to carry out the coding of a sequence of information symbols using a first code (C), of the LDPC type, defined by a first check matrix, H, or rather using a second code ({tilde over (C)}), of lesser efficiency than that of the first code, the second code being defined by a second check matrix, {tilde over (H)}, obtained by fractionation (630) of the rows of the first check matrix. The invention finds application to, among others, a link adaptation method, a turbo-coding method, a distributed coding method within a cooperative network, a packet transmission method using an IR-HARQ protocol.