Four control strategies for N-removal in alternating activated sludge plants (ASP's) are compared: 1. timer-based, 2. switching the aeration on/off when depletion of nitrate/ammonium is detected, 3. switching the aeration on/off when ammonium crosses an upper/lower-bound, 4. the newly developed adaptive receding horizon optimal controller (ARHOC) as presented in Lukasse et al. (1997). The comparison is made by simulating the controllers' application to an alternating continuously-mixed activated sludge reactor preceded by a small anoxic reactor for predenitrification. The biological processes in the reactors are modelled by the activated sludge model no. 1. Realistic influent patterns, measured at a full-scale wastewater treatment plant, are used. The results show that three totally different controllers (timer-based, NH4-bounds based and ARHOC) can achieve a more or less equal effluent quality, if tuned optimally. The difference mainly occurs in the sensitivity to suboptimal tunings. The timer-based strategy has a higher aeration demand. The sensitivity of the ARHOC controller to sub-optimal tuning, known measurement time delays and changing plant loads is significantly less than that of the other controllers. Also its tuning is more natural and explicit.