We describe an efficient method for the construction of small-insert genomic libraries enriched for highly polymorphic, simple sequence repeats. With this approach, libraries in which 40-50% of the members contain (CA)n repeats are produced, representing an approximately 50-fold enrichment over conventional small-insert genomic DNA libraries. Briefly, a genomic library with an average insert size of less than 500 base pairs was constructed in a phagemid vector. Amplification of this library in a dut ung strain of Escherichia coli allowed the recovery of the library as closed circular single-stranded DNA with uracil frequently incorporated in place of thymine. This DNA was used as a template for second-strand DNA synthesis, primed with (CA)n or (TG)n oligonucleotides, at elevated temperatures by a thermostable DNA polymerase. Transformation of this mixture into wild-type E. coli strains resulted in the recovery of primer-extended products as a consequence of the strong genetic selection against single-stranded uracil-containing DNA molecules. In this manner, a library highly enriched for the targeted microsatellite-containing clones was recovered. This approach is widely applicable and can be used to generate marker-selected libraries bearing any simple sequence repeat from cDNAs, whole genomes, single chromosomes, or more restricted chromosomal regions of interest.