The effect of feeding regimes on the performance of a laboratory-scale anaerobic ammonium oxidation (anammox) anaerobic baffled reactor (ABR) with four compartments was investigated based on nitrogen-removal efficacy, robustness, and residence time distribution (RTD) tests. The research was undertaken at 35±1°C and synthetic wastewater was used. In the normal feeding ABR (NFABR), the nitrogen-removal efficiency of the first compartment accounted for approximately 80.0% of the total nitrogen conversion, while, in the split-feeding ABR (SFABR), each compartment converted 38.0%, 30.6%, 20.0%, and 11.4% of the total nitrogen. A robustness test demonstrated that the SFABR performed steadily under moderate shock (1.5-fold) with more drastic fluctuations under 2.0- and 3.0-fold shocks compared to the NFABR. In addition, more time was needed for the SFABR to recover from the violent shocks (805min for the SFABR VS 695min for the NFABR). RTD tests implied that the flow pattern of the SFABR was similar to that of a continuous stirred tank reactor (CSTR) with greater dispersion numbers compared with the NFABR (6.11 and 3.03 versus 2.52), which may contribute to the vulnerability of the SFABR under violent shocks.