Fluoride removal by electrocoagulation/electrofloatation was investigated as an alternative defluoridation method. Batch experiments with aluminum electrodes were used to investigate, with response surface methodolgy (RSM), genetic algorithm (GA), and the artificial neural network approach (ANN), the effects on defluoridation by electrocoagulation/electrofloatation of the concentration of usual five co-existing anions (Br-, CI-, NO3-, SO42-, and PO42-), and the operational parameters of initial pH, applied electrical current intensity, initial fluoride concentration, and process time. The results demonstrated that the type and concentration of co-existing anions, as well as various operational parameters, had a significant effect on the efficiency of defluoridation by electrocoagulation/electrofloatation.