To determine the mechanism of thalidomide's antimyeloma efficacy, we studied the drug's activity in our severe combined immunodeficiency-human (SCID-hu) host system for primary human myeloma. In this model, tumor cells interact with the human microenvironment to produce typical myeloma manifestations in the hosts, including stimulation of neoangiogenesis. Because mice are not able to metabolize thalidomide efficiently, SCID-hu mice received implants of fetal human liver fragments under the renal capsule in addition to subcutaneous implants of the fetal human bone. Myeloma cell growth in these mice was similar to their growth in hosts without liver implant, as assessed by change in levels of circulating human immunoglobulins and by histologic examinations. Thalidomide given daily by peritoneal injection significantly inhibited myeloma growth in 7 of 8 experiments, each with myeloma cells from a different patient, in hosts implanted with human liver. In contrast, thalidomide exerted an antimyeloma effect only in 1 of 10 mice without liver implants. Microvessel density in the untreated controls was higher than in thalidomide-responsive hosts but not different from nonresponsive ones. Expression of vascular endothelial growth factor by myeloma cells and by other cells in the human bone, determined immunohistochemically, was not affected by thalidomide treatment in any experiment. Our study suggests that thalidomide metabolism is required for its antimyeloma efficacy. Although response to thalidomide was strongly associated with decreased microvessel density, we were unable to conclude whether reduced microvessel density is a primary result of thalidomide's antiangiogenic activity or is secondary to a lessened tumor burden.