We studied the efficiency of recurrent selection based solely on marker gen
otypes (marker-based selection), in order to increase favourable allele fre
quency at 50 previously detected quantitative trait loci (QTLs). Two select
ion procedures were investigated, using computer simulations: (1) Truncatio
n Selection (MTS), in which individuals are ranked based on marker score, a
nd best individuals are selected for recombination; and (2) QTL Complementa
tion Selection (QCS), in which individuals are selected such that their QTL
composition complements those individuals already selected. Provided QTL l
ocations are accurate, marker-based selection with a population size of 200
was very effective in rapidly increasing frequencies of favourable QTL all
eles. QCS methods were more effective than MTS for improving the mean frequ
ency and fixation of favourable QTL alleles. Marker-based selection was not
very sensitive to a reduction in population size, and appears valuable to
optimize the use of molecular markers in recurrent selection programmes.