A synthetic-fiber barrier for the removal of turbidity in water was developed and tested using a laboratory scale channel. The effects of hydraulics (flow rate and exchange rate); density current caused by temperature and turbidity difference; barrier conditions (thickness, number and shape); and particle size were analyzed. The experimental results indicated that removal efficiency was positively related to barrier thickness and number, was inversely related to the strength of the density current, and was also negatively affected by the flow rate and exchange rate. A wedged barrier was found to work better than a rectangular one when the same amount of fiber was used. Based on the experimental work, empirical models for the removal efficiency and barrier design were established using dimensionless groups. The modeling results indicated that the predicted values were consistent with the experimental work and the increases and decreases in the performance were suitably simulated.