Many threatened species are being maintained in captivity to save them from extinction, often with the eventual aim of reintroduction. The objective of genetic management in captivity is to ‘freeze’ evolution i.e. to avoid genetic adaptation to captivity and to retain genetic diversity. Most current genetic management of threatened species addresses the latter, but does not explicitly address the former. The theory underlying current genetic management and its practical implementation assumes neutrality of loci. However, genetic adaptation in captive populations may cause non-neutral behavior at neutral loci due to selective sweeps (hitchhiking) caused by rapid allele frequency changes at linked fitness loci. We compared changes in microsatellite genetic diversity at eight non-coding loci with neutral predictions in 23 pedigreed captive populations of Drosophila melanogaster maintained with effective sizes of 25 (eight replicates), 50 (6), 100 (4), 250 (3) and 500 (2) for 48 generations. Loss of microsatellite heterozygosity was significantly faster (by 12%) than predicted by neutral theory, as assessed by regressing proportion of heterozygosity retained on pedigree inbreeding coefficients. Further, greater than neutral changes were observed for both variances in allele frequencies across replicates (by 25%), and for temporal changes in allele frequencies (by 33%). All eight microsatellite loci showed signals of selectively-driven changes. Rather than having their evolution ‘frozen’, captive populations are undergoing major genomewide selective sweeps that affect not only fitness loci but linked neutral loci. Captive genetic management for threatened species destined for reintroduction requires modification to explicitly minimize genetic adaptation to captivity.