The effect of shear on the formation and break-up of floes generated using aluminum sulfate ('alum'), polyaluminum chloride and two cationic polyelectrolytes has been investigated using conventional jar test procedure and by continuous optical monitoring. Using the experimentally determined optimum dosage, the breakage of floes was followed at a high stirring speed, corresponding to an average shear rate (G) of about 520 s(-1). Most of the breakage occurred within a few seconds of increasing the shear rate. After each breakage the stirring rate was reduced to the original value of 50 rpm (G approximate to 23 s(-1)) to allow the floes to re-grow. For alum and polyaluminum chloride, it was found that only limited re-growth of floes occurred indicating a significant irreversibility of the floc break-up process. Residual turbidity increased after floc breakage and re-growth, indicating a reduction in sedimentation rate of the re-grown flocs, consistent with the continuous monitoring results. For cationic polyelectrolytes, the re-growth of floes occurred to a much greater extent in one case where floc breakage was almost fully reversible. Similar residual turbidity values before floc breakage and after re-formation were found, supporting the reversibility of floc breakage with cationic polyelectrolytes. The effect of rapid stirring (breakage) time was also examined. Floes were exposed to the rapid stirring rate for different periods (5 - 300 s). Floc strength and recovery factors were found to decrease with increased breakage time for most of the coagulants tested. It was also found that the floc size tends to a nearly constant value after an initial abrupt fragmentation, with only a very slow size reduction over several minutes. (C) 2004 Elsevier B.V. All rights reserved.