The kinetics of 131I-β2-microglobulin (β2-M) were studied using external total body gamma counting in a low noise chamber after administration of trace doses of radioactivity (4 μCi) in 14 uremic patients treated by either hemodialysis or hemofiltration. Data were collected over a 1 week period that included 3 dialysis sessions. The following artificial membranes were used: Cuprophan, polyacrylonitrile AN69, polysulfone, polymethylmethacrylate (PMMA), and polyamide. Radiolabeled β2-M excretion by an extrarenal route was nearly nonexistent. The 131I-β2-M half-life was between 2.4 and 8 days, shorter in patients with residual diuresis. A mean removal of 153 ± 33 mg/L of β2-M was obtained per dialysis session with a highly permeable membrane. A hemofiltration session (25 L exchange per session) was slightly more efficient in removing β2-M than a 4 h hemodialysis session with the same AN69 highly permeable membrane. The amounts of 131I-β2-M binding on the membranes, expressed as β2-M equivalents, were 0, 16, 54, 58, and 59 mg/m2 for Cuprophan, polysulfone, polyacrylonitrile AN69, polyamide, and PMMA, respectively. In conclusion, the decrease of total body gamma counting directly reflected the β2-M breakdown and removal in hemodialysis patients. Intact β2-M was removed by convection with synthetic, highly permeable membranes. In addition, membrane adsorption accounted for 15% (polysulfone) to near 100% (PMMA) of the β2-M removal per session. Adsorption was of the same magnitude regardless of the dialysis technique in use, indicating a membrane saturability process. None of the currently available dialysis procedures based on a 3 sessions per week schedule can balance β2-M generation. [ABSTRACT FROM AUTHOR]