The efficiency of marker-assisted introgression programmes, expressed as genetic lag relative to a commercial population under continuous selection, was investigated using analytical methods. A genetic model was assumed for which the genetic variance in the introgression population was a function of the within-breed genetic variance and the initial breed difference. It was found that most of the genetic lag occurs in the latter stages of an introgression programme, when males and females which are heterozygous for the alíele to be introgressed are mated to produce homozygous individuals. Reducing genetic lag through selection on genomie proportion by using genetic markers throughout the genome, i.e. by selecting heterozygous individuals which resemble the recipient (commercial) population most, was effective if the initial breed difference was very large (e.g. 20 within-breed phenotypic standard deviations). In that case, selection solely on genetic markers could be practised to speed up genome recovery of the commercial line. If the initial breed difference is small, phenotypic or best linear unbiased prediction (BLUP) selection is superior in reducing genetic lag under the assumed genetic model.