Aerobic granular sludge technology offer a possibility to design compact wastewater treatment plant based on simultaneous COD, nitrogen and phosphate removal in a sequencing batch reactor.Simultaneous nutriment removal was possible, because of heterotrophic growth inside the granules (denitrifying and PAO organisms). Selection for slow growing organisms such as PAO organisms (byincrease of the feeding period time) also allow to improve granule stability and density. This study was aimed to investigate the granule formation under aerobic conditions and to gain insightinto the mechanisms responsible for aerobic granulation. The study was conducted in two bubblecolumn reactors the only difference was feeding mode. In the first reactor pulse feeding was applied and in the second reactor long term feeding under anaerobic condition was applied. The temperature inthe reactor was not regullated and depend on room temperature (30 ± 5 °C) and humidity. In the anaerobic feeding reactor, smooth granules were obtained in a first period, then they become fluffy and washout from the reactor. Increases of shear lead to recover the system. In about two weeks, smooth and dense granules were obtained.During the experiment granules were unstable, theirs density vary between 46 and 56 g VSS/l of granulesand their average size fluctuated between 1.1 and 2.3 mm. Several time biomass washouts occur for unknown reasons; the reactor never reaches the steady state condition. Nitrogen and phosphate removal were not occur, consequently, its lead to the conclusion thatdenitrifying and PAO organisms were not present in system. Granules instability could be explained by the absence of these organisms and by the temperature fluctuations. In pulse feeding reactor, due to a pH accident, granules disappear letting place to fluffy organisms. Those fluffy organisms reappear in the system after a complete wash and restart of the reactor. Fluffyorganism out compete proper granules and the flocs size became considerably high. The fluffy organisms present in both reactors seem to be yeast like fungi: Geotrichum sp.