Online hemodiafiltration (OL-HDF) has now demonstrated some benefits in reducing mortality. It seems that rising convective volumes improve the outcomes, but the risks of it, such as albumin leakage, are not well defined yet. The aim of the present study was to evaluate the albumin leakage using two different filters with 20 and 30 L of post-dilution OL-HDF. In this cross-sectional study, 20 prevalent patients receiving post-dilution OL-HDL were included. We analyzed two dialyzers: FX1000, FMC and Polyflux 210H, Gambro. During four consecutive dialysis sessions, monitors were programmed using control-volume to obtain 20 or 30 L with both dialyzers. We collected albumin samples of the effluent at 5, 15, 30, 45 and 60 min and performed area under the curve (AUC) determinations for evaluating the losses. Mean patient age was 60 ± 9 years, and 70% were men. Albumin leakage was significant higher with Polyflux 210H when compared to FX 1000 FMC. A convective volume of 30 L produced greater albumin leakage than 20 L with both filters, though only with the FX 1000 FMC was it significant (minimum albumin leakage during first hour with FX 1000 FMC 20 L: 79.2 [0.0–175.7] mg; 30 liters: 403.3 [63.5–960.7] mg; with PF 210 Gambro 20 L: 869.1 [420.0–3214.7] mg; 30 L: 1841.7 [443.8–3417.5] mg). During OL-HDF, convective transport causes albumin leakage at least during the first hour. The albumin concentration in the effluent differs according to the type of filter used and the convective volume.