Adoptive cell therapy (ACT) using tumor-infiltrating lymphocytes (TIL) for metastatic melanoma has shown objective response rates as high as 72%. The successful application of this therapy requires the selection of unique tumor-reactive lymphocyte cultures for each patient. This is a technically and logistically difficult undertaking, and patients who do not have tumor-reactive TIL are not considered eligible for treatment. To simplify the methods of TIL generation and extend TIL-based immunotherapy to additional patients, methods were developed to use unselected, minimally cultured ("young") TIL. Young TIL cultures contain a variable number of CD8(+), CD4(+), and CD3(-)CD56(+) natural killer cells. In this study we retrospectively investigated a role for these subsets in the clinical outcome of patients treated with TIL derived from selected microcultures. This analysis demonstrated a suggestive but nonsignificant association between the number of CD8(+) cells administered and tumor regression. We therefore investigated the feasibility of selecting CD8(+) cells from young TIL cultures for ACT therapy. The available methods for clinical scale CD8(+) enrichment proved inadequate for TIL, so an optimized CD8(+) enrichment method was developed and is reported here. We observed that CD8 (+)enrichment of some TIL cultures revealed in vitro tumor recognition that was not evident in bulk culture, and an improved in vitro recognition of tumor in other TIL cultures. In addition, the enriched CD8(+) young TIL expanded more reliably and predictably in rapid expansions than the bulk TIL. Thus, optimized CD8(+) selection combines the benefits of antigen-selected TIL and young TIL for generating lymphocyte cultures for ACT, and should be evaluated in cell transfer therapy protocols.