Electric split-flow thin (SPLITT) fractionation permits the continuous separation of charged species, particularly proteins, at gram and subgram levels. We characterized this system and separated protein mixtures based on the difference between protein isoelectric points (pi). For characterization, we examined seven variables. Buffer stability was determined by measuring pH changes per hour and the UV spectrum before and after an electrical potential of 50 V was applied. The electrical field across the channel was determined by measuring he buffer conductivity and the current passed through it. The experimental and theoretical (calculated) fractional retrieval of proteins was determined by the relative magnitude of field-induced and outlet flow rates. The protein response at various electrical fields (0, 10, 20, 30 V) and solution pHs (4.85, 5.60, 6.87, and 7.80) was examined, as were the effects of the ionic strength of the buffer, protein recovery, and protein separation with pulsed sample injection. To separate protein mixtures after the system was characterized, we ran continuous SPLITT fractionation of five protein mixtures for more than 8 hours. Characterization results show that 1) buffer stability was good for acetate and phosphate buffers, 2) the electrical field across the channel was about 60% of that predicted by a geometric estimation, 3) experimental retrieval of four proteins (ferritin, BSA, hemoglobin, and cytochrome c) agreed well with calculated retrieval, 4) protein response at the four electrical fields and four solution pi-Is corresponded to the difference between protein pi and solution pH, 5) lower buffer ionic strength was better for protein separation, 6) protein sample recovery was reasonable from 78 to 90% (mean 85%) for six proteins, and 7) pulsed sample injection led to successful separation of five protein mixtures. In the second part of the study, three protein mixtures were successfully separated using continuous Separation over 8 hours. The collected fractions showed clean separation as confirmed by flow field-flow fractionation and spectrophotometer analysis. The throughput was around 15 mg/h and the minimum difference between protein pIs that permitted separation was about two units. We conclude that isoelectric SPLITT fractionation has potential for use in protein purification.