The present investigation emphasizes on the biosorptive removal of toxic pentavalent arsenic from water using steam activated carbon prepared from mung bean husk (SAC-MBH). Characterization of the synthesized sorbent was done using different instrumental techniques, i.e., SEM, BET and point of zero charge. Sorptive uptake of As(V) over steam activated MBH as a function of pH (3-9), agitation speed (40-200 rpm), dosage (50-1000 mg) and temperature (298-313 K) was studied by batch process at arsenic concentration of 2 mg L -1 . Lower pH increases the arsenic removal over the pH range of 3-9. Among three adsorption isotherm models examined, Langmuir model was observed to show superior results over Freundlich model. The mean sorption energy (E) estimated by Dubinin-Radushkevich model suggested that the process of adsorption was chemisorption. Thermodynamic parameters confer that the sorption process was spontaneous, exothermic and feasible in nature. The pseudo-second-order rate kinetics of arsenic gave better correlation coefficients as compared to pseudo-first-order kinetics equation. Three process parameters, viz. adsorbent dosage, agitation speed and pH were opted for optimizing As(V) elimination using central composite design matrix of response surface methodology (RSM). The identical design setup was used for artificial neural network (ANN) for comparing its prediction capability with RSM towards As(V) removal. Maximum arsenic removal was observed to be 98.75% at sorbent dosage 0.75 gm L -1 , pH 3.0, agitation speed 160 rpm and temperature 308 K. The study concluded that SAC-MBH could be a competent adsorbent for As(V) removal and ANN model was better in arsenic removal predictability results than RSM model.