We compare repurchase costs from discriminatory price vs. uniform price mechanisms in the Korean treasury bond repurchase auction market. For this purpose, we employ detailed bidder level micro data for each of 20 discriminatory repurchase auctions carried out in the Korean government bond market in 2011. We first theoretically recover unobserved individual bidding function under the “counter-factual” uniform price repurchase auction from the observed bidding function under the actual discriminatory repurchase auction, and then empirically estimate the repurchase cost differences between the two auction mechanisms. To test significance of the differences, we use Bootstrap re-sampling methods where uncertainty in the cut-off yield spreads and uncertainty in the participating bidders are addressed individually as well as jointly. Our results indicate that the uniform price repurchase auction decreases the repurchase cost relative to the discriminatory one in each of the 20 cases analyzed in this paper.