The presence of excess fluoride in drinking water has become a potential cause of agony to hundreds of million people the world over as it initiates the debilitating disease fluorosis. This concern imparts enough impetus among the water community for concerted research on defluoridation of drinking water. The present study evaluates the feasibility of using an adsorbent, alumina cement granules (ALC) in removing fluoride from water through batch and column studies. In batch studies, it was observed that a dose of 2 g l(-1) of ALC could bring down fluoride concentrations in water from 8.65 mg l(-1) 1 to below the permissible level of 1.0 mg l(-1) at optimum conditions. The equilibrium sorption data, generated by dose variations of ALC, were better modeled by Freundlich isotherm. The maximum monolayer capacity of ALC suggested by the Langmuir model was 10.215 mg g(-1). In column studies, the maximum adsorption capacity of ALC at the point of breakthrough was found to be 2.27 mg g(-1) at a flow rate of 4 ml min(-1). The responses of the adsorbent in the fixed bed for varying operating conditions of bed depth, flow rate, and initial fluoride concentrations were analyzed. The sorption kinetic models; Hutchins BDST, Thomas, Yoon-Nelson, and Clark were examined to describe the sorption process. The characteristic parameters of the respective models for the process design of columns were obtained by their linear regression. Among all the models tested, the Clark's model could better describe the sorption process at all sorptive conditions and ranges analyzed. The study also suggested the use of sorption contours of effluent fluoride concentration for better visual evaluation and correlation of experimental sorption profiles with the model. The sorption contours could aid in the quantitative evaluation of service times of columns at various stages of breakthrough for any bed depth, flow rate and initial fluoride concentration within the ranges studied. (C) 2007 Elsevier B.V. All rights reserved.