A bench-scale short-term test, developed to predict the long-term integrity of membranes with potential for use in anaerobic fluidized-bed membrane bioreactors, was used to evaluate several commercial hollow-fiber membranes. It was found that membrane performance varied widely, some membranes failing much more rapidly than others. Also found was that larger sizes of the fluidized media, in this case granular activated carbon (GAC), severely affected membrane structural integrity more than did smaller sizes, as did the method used for membrane attachment. Within the limits studied, the GAC packing ratio had only a minor impact. A decrease in membrane permeability that sometimes resulted during the testing and was caused by the deposition of fine GAC particles could be eliminated without membrane damage through simultaneous chemical cleaning and sonication. This new testing procedure should be useful for selecting membranes and reactor operating conditions to better ensure long-term operating performance of anaerobic fluidized-bed membrane bioreactors.