Toxin kinetic modeling (TKM) for hemodialysis patients is an established method to understand the physiological distribution of toxins, their accumulation, and removal characteristics. To estimate unknown parameters in TKMs, patient blood samples are collected during and after dialysis. The question is when and how many samples should be collected? Among existing clinical studies, there is no definite consensus on this issue. We employed the model-based design of experiments technique to elucidate the optimal sampling protocol. As the name suggests, its pre-requisite is a model we considered the diffusion-adjusted regional blood flow model for beta(2)-microglobulin. This model comprises three parameters. A total of 12 scenarios corresponding to 3 experiment durations (240, 300, or 360 min) x 4 sampling regimes (7, 9, 11, or 13 samples) are simulated, where 240 min correspond to conventional dialysis duration. For each scenario, the optimal experiment is designed; parameters are estimated, and compared with known true parameters. The 300 min experiment (i.e. 60 min post-dialytic wait) with 11 samples is considered optimum among all optimal experiments owing to comparable point estimates, fewer samples, and shorter waiting time after dialysis. Contrary to existing sampling protocols in literature, parameters are better estimated using samples collected in the intra-dialytic phase. The samples in the post-dialytic phase are less informative, and should be collected only towards the end of the post-dialytic phase. (C) 2015 Elsevier Ltd. All rights reserved.