Abstract North West Water Ltd (NWW) has applied hollow-fibre membrane microfiltration (MF) for treatment of potable water at full scale (15–80 MLD). NWW has currently five MF plants either in operation, commissioning or under construction (total capacity 240 Ml/d). The adoption of this technology has been driven by a number of factors, one of which is to provide a barrier to Cryptosporidium . As part of the development of large systems, NWW's Technology Development Team have performed an extensive series of pilot trialling. The efficacy of the backwash is one of the key criteria in ensuring the efficient and effective operation of the MF system. This is particularly important with out-to-in hollow-fibre membrane systems, since the bundle of fibres can act as a depth filter and accumulate deposits from the feed water. Moreover, the internal membrane hydraulics of an out-to-in system, during backwash, is complex. This makes optimisation of the efficient removal of solids from the membrane surface, by the backwash, an important task. Poor flow distribution, within the membrane module, can lead to the development of areas of low flow rates or “dead areas”. There is a risk of solids accumulation, within these dead areas, which will lead to increased transmembrane pressure during production. Pilot plant trials have been performed to optimise the backwash conditions at all the sites at which NWW have applied MF. The work has explored a variety of parameters associated with the backwash operation, such as backwash pressure, interval, duration, volume, chemical assistance, fibre integrity and the use of water pulsing. Results show that all the variables must be addressed to develop an optimised backwash strategy. By optimising backwash pressure, interval, duration, volume and use of chemicals (oxidants) run times can be extended and hence operating costs controlled. The most promising results, in terms of removal of deposited foulant from the membrane surface, are obtained with the pulsed backwash, where the backwash water is delivered in a series of discrete pulses. These studies have clearly demonstrated that the majority of deposited material is detached from the membrane surface during the initial phase of the backwash. This suggests that the detachment mechanism is more likely to be mechanical, caused by the sudden expansion of the bundle, and not hydraulic shear. Optimisation of pulse duration, the interval between pulses and the number of pulses applied in any one backwash cycle, has been performed. It has been possible to show that after time a steady state is achieved and backwash efficiency approaches 100%. Interestingly, backwash efficiency was found to increase as a run progressed. At the outset of a run backwash efficiency was typically low (∼14%) and increased with time to nearly 100%. This indicates that some form of steady state system is in operation whereby the solids being removed from the system are a function of the incoming solids and the previously retained solids. The effect of fibre breakage on backwash efficiency, with an out-to-in membrane system was found to result in serious degradation of backwash efficiency. The likely explanation is that a broken fibre will take a significant proportion of the backwash flow. This effect was observed during pilot plant trials when backwash efficiency and permeate water quality deteriorated simultaneously. Subsequent bubble point tests revealed fibre breakage. Once this damage was repaired. backwash efficiency was restored.