We thank Badoual et al for their correspondence regarding our study, in which we found that a high density of intratumoral FOXP3 regulatory T cells (Tregs) in colorectal cancer (CRC) was associated with good prognosis. Although our findings were in contrast with observations published for several other solid tumor types, Badoual et al argued that our results should not have been entirely unexpected in light of previous results obtained in animal models of colon cancer and in human lymphomas and head and neck cancers. They also proposed some appealing explanations of why Tregs may be linked to better survival, including direct antitumor effects, thus providing a stimulus for additional investigation in this field. Two preclinical studies in which the adoptive transfer of Tregs in mouse models was shown to prevent the formation of adenomas and cause regression of established tumors were cited by Badoual et al as being supportive of our findings. However, we evaluated Treg densities in tumor and adjacent normal mucosal tissues and have no data on how these parameters relate to the level of circulating Tregs. Clearly, this would be an interesting area for additional study, particularly in relation to possible changes in blood and tissue Tregs caused by cytotoxic chemotherapy and whether this affects the antitumor immune response. The contrasting prognostic associations observed in our study for Tregs localized within tumor and adjacent normal tissue highlight the need for investigation of both compartments, perhaps also in conjunction with the level of circulating Tregs. A different prognostic role for Tregs according to tumor stage was one explanation put forward by Badoual et al to explain contradictory results in the literature. Although our study did not examine metastatic CRC cases, we found that high Treg density was associated with better survival in both stage II and stage III CRC. Second, CD8 to FOXP3 and CD45RO to FOXP3 ratios were examined for prognostic significance but found to be inferior to FOXP3 cell density alone. However, we agree that information on the relationship between FOXP3 cell density and the expression of effector molecules such as granzyme and perforin would be useful; indeed, this is currently being investigated by our group. As discussed by Badoual et al, functional studies on tumor-infiltrating FOXP3 cells are required to help explain the biologic basis for their association with prognosis in CRC and in other cancer types. Regardless of why the density of tumor-infiltrating Tregs appears to have different prognostic value for different cancer types, the fact remains that it was a strong and independent marker in our study of CRC. Confirmation of our findings by independent studies could lead to the routine use of Treg density, in combination with vascular and serosal invasion, for the improved prognostic stratification of stage II CRC.