A method of optimizing the size of blocks of coded data intended to be subjected to an iterative decoding process, a maximum error rate of the iterative decoding process being fixed in advance, in which there are sought, among a plurality of block sizes (N/k) which are submultiples of the normal block size by an integer factor (k) greater than or equal to 1 and a plurality of integers giving the maximum number of iterations that can be effected by the said iterative decoding on a block, (1) a submultiple size, and (2) a maximum number of iterations such that they are compatible with the maximum error rate, and such that the mean number of iterations that will be applied by the iterative decoding process on a block of submultiple size is minimized.