The overall objective of the present study was to characterize the effect of operating a submerged air sparged membrane system over a wide range of operating conditions (i.e. sub- and super-critical flux conditions) on the extent and mechanisms of membrane fouling during drinking water treatment. The sub- and super-critical flux conditions considered were generated by varying the operating permeate flux, the bulk cross-flow velocity, the flow characteristics (i.e. with or without air sparging) and the configuration of the membrane modules. Regardless of the operating conditions, the increase in the trans-membrane pressure during operation could be effectively modeled using a simple exponential relationship of the form P-v = P(0)e(KV) where P-v was the trans-membrane pressure after filtering a given volume of permeate V, PO was the initial trans-membrane pressure, and K was the overall fouling coefficient. The type and extent of fouling could be characterized into three operating zones based on a flux index. At a relatively low flux index, a 'true' sub-critical flux zone was observed in which the overall fouling coefficient was constant for the different experimental conditions examined. For operation in this zone, fouling appeared to be predominantly due to internal fouling. At a flux index greater than zero, a super-critical flux zone was observed in which a linear relationship was observed between the overall fouling coefficient and the flux index. For operation in this zone, fouling appeared to be predominantly due to surface fouling. A transition zone was observed between 'true' sub-critical flux conditions and super-critical flux conditions, in which fouling appeared to be increasingly due to surface fouling. The overall fouling coefficient, as well as the evolution of the trans-membrane pressure in a submerged air sparged hollow fiber membrane system, could be effectively modeled for all operating zones using a relatively simple semi-empirical relationship which considered the back-transport of particles (i.e. foulants) from the membrane surface. It is expected that this simple relationship can be used in parallel with pilot-scale testing and reduce the extent of testing needed to identify the parameters that minimize fouling. (C) 2007 Elsevier B.V. All rights reserved.