The correction of mutant beta-globin genes has long been a therapeutic goal for patients with beta-thalassemia or hemoglobinopathies. The use of homologous recombination (HR) to achieve this goal is an attractive approach because it eliminates the need to include regulatory sequences in the therapeutic construct, and it eliminates mutagenesis induced by random integration. However, HR is a very inefficient process for gene correction, and its efficiency is probably locus dependent. The length of targeting arms is thought to be a determinant of targeting efficiency, so we compared the ability of standard (8-kb) versus very long (16-, 24-, and 110-kb) regions of homology to correct a mutant murine beta-globin gene in embryonic stem cells. Increasing the length of the targeting sequences did not increase the efficiency of HR in this locus, suggesting that alternative approaches will be required to improve the efficiency of this approach for globin gene correction.