A pilot study for grafting of patients with vitiligo using cultured epithelial autografts containing melanocytes gave disappointing clinical results, with pigmentation achieved in only one out of five patients. Irrespective of the fate of melanocytes grafted back onto the patients, we experienced problems in identifying melanocytes within these well-integrated keratinocyte sheets. This led us to explore the fate of these cells within these sheets in vitro and to seek to improve their number and function within the sheets. We report that the introduction of a fibroblast feeder layer can improve melanocyte number within melanocyte/keratinocyte co-cultures initially, but at very high keratinocyte density, there is a marked loss of melanocytes (as detected by staining for S100). Additionally, we found that keratinocytes not only down-regulate melanocyte number, but also pigmentary function; thus, it was possible to identify melanocytes that were S100 positive but tyrosinase-related protein-1 (TRP-1) negative in confluent well-integrated keratinocyte sheets. In summary, our data suggest that keratinocytes at high density initially suppress melanocyte pigmentation (as evidenced by a lack of TRP-1 expression) and then cause a physical loss of melanocytes. The introduction of a fibroblast feeder layer can help maintain melanocyte number while keratinocytes are subconfluent, but fails to oppose the inhibitory influence of the keratinocytes on melanocyte TRP-1 expression.