The effect of tapered-shear (diminishing velocity gradient) flocculation on the formation, breakage and reformation of flocs using aluminium sulphate and an anionic polymer were evaluated. An on-line continuous optical monitoring technique was used for this purpose. Two different mean shear rates, G=60s−1 and G=100s−1 were tried in the slow stirring stage of the experiments. For the mean G=60 sec−1, sample was stirred at 400 rpm (G=518 sec−1) for 10 seconds then the stirring speed was reduced to 100 rpm (G=60 s−1) and held at this value for the required time (9 minutes). In the other trial, slow stirring speeds of 130 rpm, 100 rpm and 70 rpm and 150 rpm, 100 rpm and 50 rpm for 3 minutes each were used. After slow stirring, the speed increased to 400 rpm (G=518 s−1) for 10 seconds and then reduced back to appropriate slow stirring speed as stated above. A similar approach was applied for mean G=100 s−1. For continuous tests, a laboratory scale set-up was constructed as a tapered flocculation system that consists of three chambers. Flocs formation in the first and third chambers were monitored and compared. Larger flocs that have higher reformation rate were produced with tapered flocculation.