The EBV-producing marmoset B-cell line (B95-8), commonly used as a source of EBV for stimulation and transformation of human B cells, was shown to proliferate in response to supernatants containing human B-cell growth factors (BCGF) derived from PHA-activated T cells or the KG-la cell line, and to a commercial low molecular weight BCGF (BCGFlow), but not to recombinant human IL-4 (rhIL-4). In this respect, B95-8 responded in much the same way as human EBV-transformed lymphoblastoid cell lines (LCL). In contrast, B95-8 did not secrete immunoglobulin in response to B-cell differentiation factor (BCDF) containing supernatants from the KG-la cell line, nor to BCGFlow, or IL-6 obtained from the T24 bladder carcinoma cell line, whereas significant responses were obtained with human EBV-transformed LCL. Both B95-8 and control EBV-transformed human LCL secreted BCGF and BCDF detected with the indicator B-cell lines CESS, L4, and HFB1, but only the human LCL secreted BCGF detectable in co-stimulation assays with TPA-activated tonsillar B cells. Unlike EBV-transformed LCL, B95-8 did not express detectable surface CD23, and did not release into the culture medium soluble CD23 (sCD23) recognized by an EIA for the human molecule. Although not releasing detectable sCD23, B95-8 cells did proliferate in response to purified human sCD23, and were found to be 1000 times more sensitive in this assay than EBV-transformed LCL. This may provide a basis for a sensitive bioassay for sCD23. Unlike EBV-transformed LCL, it seems that in vitro proliferation of B95-8 may involve an autocrine loop which does not depend on CD23.