In recent studies, semi-selective compared to antigen-specific immunoadsorption (IA) columns showed comparable effectiveness in anti-A/B antibody removal before incompatible living donor kidney transplantation. Semi-selective columns allow a greater number of IA treatments at lower costs. They are also capable of removing potentially harmful human leukocyte antigen alloantibodies. Nevertheless, additional plasma exchange treatments are often necessary to reach the preoperative target titer, most likely due to an inadequate anti-A/B IgM antibody depletion. We compared the effectiveness of immunoglobulin and anti-A/B antibody reduction by different semi-selective (Therasorb Ig-flex, Therasorb Ig-omni5, Immunosorba) and antigen-specific (Glycosorb) IA columns during the desensitization of 63 ABO-incompatible living donor kidney transplant candidates with a total of 375 IA treatments. Fifty-three patients were eventually transplanted. Total IgM reduction during the first IA treatment was significantly greater with the Therasorb Ig-omni5 compared to the Therasorb Ig-flex (mean: -71.3 vs -41.6; p = 0.001) or Immunosorba columns (mean: -71.3 vs -42.8; p = 0.03). During a median of 5.5-6 pre-transplant IA treatments, Therasorb Ig-flex and Therasorb Ig-omni5 columns were equally effective in the reduction of total IgM while both showed superior IgM reduction compared to the Immunosorba columns (Therasorb Ig-flex, mean: -81.2 vs -72.2; p = 0.01; Therasorb Ig-omni5, mean: -88.2 vs -72.2; p = 0.02). IgG reduction was not significantly different between groups. Likewise, anti-A/B IgM antibody reduction (titer Saline) during the first IA treatment was significantly greater with the Therasorb Ig-omni5 compared to the Therasorb Ig-flex (mean titer reduction: -1.9 vs -1.1; p = 0.02) and tended to be greater than with Immunosorba or Glycosorb columns. During a median of 5-6 pre-transplant IA treatments, overall anti-A/B IgM antibody reduction was significantly greater when IA was performed with the Therasorb Ig-flex (mean titer reduction: -3.8 vs -1.3; p < 0.001) or Therasorb Ig-omni5 (mean titer reduction: -4.3 vs -1.3; p = 0.01) compared to the Immunosorba columns with no differences compared to the Glycosorb columns. Again, anti-A/B IgG antibody reduction (titer Coombs) was not significantly different between groups. The semi-selective Therasorb Ig-omni5 device offers potential advantages in reducing total IgM as well as anti-A/B IgM antibodies. Copyright © 2019 Elsevier B.V. All rights reserved.