Introduction. Acute renal failure (ARF) develops in 10–15% of patients with multiple myeloma (MM). Typically, complexes of serum free light chains (sFLC) and Tamm-Horsfall protein form casts in the distal tubules that block urine flow. Many patients, subsequently, require long-term hemodialysis (HD) and may have reduced survival. In an attempt to minimise renal damage, plasma exchange (PE) has been used to reduce the pre-renal load of sFLC. Zucchelli et al., (Kidney Int 1988; 33:1175–1180) showed that PE reduced HD requirements but subsequent controlled trials showed no benefit (Johnson WJ et al., Arch Intern Med 1990; 150:863–869: Clarke WF et al., Haematologica 2005; 90(s1) p117). The availability of sFLC immunoassays now allows an informed evaluation of the role of PE and/or HD in treating these patients.Methods. sFLC were measured in 1: 100 patients with chronic renal failure (CRF), GFR< 15ml/min and not on dialysis; 2: 38 at the beginning and end of HD (dialyzer A) and 3: 25 on chronic ambulatory peritoneal dialysis (CAPD). sFLC hemofiltration efficiency was assessed, in-vitro, for three different dialyzers, using standard pump pressures and flow rates. Sera containing approximately 1,000 mg/L of monoclonal kappa (κ ) (NR < 20mg/L) and lambda (λ ) FLC (NR < 27mg/L) were recycled for ~45 minutes through the dialyzers and clearance rates assessed. A two-compartment, mathematical model was constructed to assess sFLC removal by HD and PE from hypothetical patients. The following parameters were considered:- sFLC concentrations in MM at clinical presentation; monomeric κ or dimeric λ clearance differences; partition of sFLC between vascular and extravascular compartments; flow of FLC between compartments; half-life of sFLC in ARF; sFLC production and tumour killing rates with chemotherapy. The model was interrogated for various dialysis times, different dialyzers and a PE protocol of 3L, x6 over 2 weeks.Results. Mean sFLC concentrations in severe CRF were: κ 93 mg/L (range 43–207); λ 64 mg/L (range 43–134). Mean sFLC before and after HD: κ 130 mg/L (range 40–567) to 49 (range 20–234); λ 100 mg/L (range 31–225) to 61 (range 24–159). Mean sFLC in CAPD patients: κ 118 mg/L (range 31–266); λ 114 mg/L (range 36–263). Clearances of sFLC by the 3 dialyzers were: A: BBraun high flux polyethersulfone 1.8sqm (HI PeS) (the dialyzer in routine use); κ 46%: λ 39%, B: Idmesa high flux polyethersulphone 2.0sqm (200MHP); κ 67%: λ 59%, C: Asahi high flux polysulphone 2.1sqm (APS 1050); κ 71%: λ 65%. Model calculations showed that sFLC reduced from a starting value of 14g/L (typical of light chain MM) to less than 0.5g/L in 14 days using PE. Membrane A, used for four hours, x3/week, was approximately 20% more efficient than PE and reduced κ sFLC approximately 50% faster than the natural clearance rate of patients in ARF. Dialyzers B and C were approximately twice as efficient. Performing initial, 16-hour dialyses daily for 3 days, with dialyzer C reduced sFLC to less than 0.5g/L in 2–3 days with ~95% of the sFLC being removed. Dimeric λ sFLC were removed ~50% more slowly.Conclusion. PE was less efficient at removing sFLC than routine HD. Prolonged HD with high-flux dialyzers removed monoclonal sFLC quickly and should be assessed in a clinical trial for patients with acute myeloma kidney. CAPD was inefficient at removing sFLC.