Conventional jar tests and on-line size monitoring were used to investigate the effects of slow-mixing intensity and duration on residual turbidity and floc size during charge neutralization coagulation and sweep flocculation with polyaluminum chloride. The compensatory effect of slow-mixing on coagulation performance following inadequate or excessive rapid-mixing was also examined. It is found that slow-mixing intensity has a more marked positive effect on charge neutralization coagulation than on sweep flocculation. The optimal root-mean-square velocity gradient, G, for slow-mixing is 15 s(-1) for both coagulation mechanisms, and charge neutralization coagulation requires a longer slow-mixing duration. The optimal slow-mixing duration, based on residual turbidity, is longer than the time to form the largest mean flocs. The optimal product of G and mixing duration, GT, for slow-mixing during charge neutralization coagulation (13500) are higher than that during sweep flocculation (4500) and both are less than the range of values recommended by the American Water Works Association (24000-84000). The optimal GT value under various slow-mixing conditions increases with G. Appropriate extension of slow-mixing duration during charge neutralization coagulation can improve coagulation performance after an inadequate or excessive rapid-mixing duration, but during sweep flocculation, appropriate shortening of slow-mixing duration after an excessive rapid-mixing or appropriate extension of slow-mixing duration after an inadequate rapid-mixing is favorable.