BACKGROUND Pregnancy-limited antiretroviral therapy (PLAT) drastically reduces HIV-1 transmission to the newborn, but may select for antiretroviral drug resistance mutations in mothers. METHODS We evaluated antiretroviral-naive, HIV-1-infected pregnant women who received PLAT between 1998 and 2005, and had 2-month or 6-month postpartum plasma samples available with HIV-1 RNA levels more than 500 copies/ml. Postpartum drug resistance mutation rates were assessed blindly using population sequencing and allele-specific PCR (ASPCR) of the M184V, K103N and D30N mutations. Factors associated with selection of drug resistance mutations were investigated. RESULTS One hundred and forty-six women were included. All women received zidovudine and lamivudine during pregnancy; 76% also received nelfinavir and 8.2% nevirapine. Resistance data were available from 114 women (78%). Postpartum rates of single-class, dual-class, and triple-class resistance were, respectively, 43, 6.1 and 0% (63.2, 10.5 and 1.7% by ASPCR). In women receiving dual or triple PLAT, respectively, postpartum M184V/I rates were 65% (95% by ASPCR) and 28.7% (51.6% by ASPCR), respectively (P < 0.01). Postpartum nonnucleoside reverse transcriptase inhibitor (NNRTI) resistance rates among women receiving nevirapine were 25% for K103N (37.5% by ASPCR) and 12.5% for Y188C. Protease inhibitor resistance rates in women receiving nelfinavir were 1.1% for D30N (1.1% by ASPCR) and 1.1% for L90M. Dual versus triple PLAT and prolonged zidovudine exposure were associated with selection of M184V. Nevirapine use and length of zidovudine and lamivudine exposure were associated with selection of K103N. CONCLUSIONS PLAT is associated with frequent selection of resistance to drugs with low-genetic barrier. Triple-drug PLAT decreases the odds for M184V selection. Routine postpartum genotypic resistance testing may be useful to guide future treatment decisions in mothers.