Adsorption kinetics involved in the removal of phenol molecules polluting waterbodies was scrutinized using the popular and rigorous Homogeneous Surface Diffusion Model (HSDM). Adsorption of phenol depends on its molecular properties, characteristics of the adsorbent and operating conditions (Liu et al., 2010). These affect the dynamic and equilibrium parameters of the HSDM. Due to a boundary condition that involves an integral and a nonlinear adsorption isotherm, an analytical solution is not possible. The parameters of the HSDM, viz. the convective mass transfer coefficient in the fluid phase and diffusion coefficient in the solid phase are difficult to estimate from experimental kinetics data, through conventional optimization algorithms. These algorithms are sensitive to the quality of initial guesses and may frequently stall if the error function surface has a nearly flat topology. A novel, initial guess free and robust algorithm, termed as Elephant Herd Optimization algorithm (EHO), was developed in-house for reliably estimating the HSDM parameters. EHO is based on the social behavior of a herd of elephants and their instinct to navigate to water sources based on their strong memory. The proposed algorithm was considerably improved from the well-known particle swarm optimization technique (Kennedy and Eberhart, 1995). Conventional methods such as Nelder-Mead, genetic algorithm, and simulated annealing were compared against EHO for the attainment of global minimum and computational efficiency. EHO was successful in identifying the parameters with the smallest attainable error sum of squares based on experimental and predicted concentration values, lesser computational effort, lesser number of function evaluations and greater precision. EHO typically reduced the time required for parameter estimations by about 30% on an average with 40% lesser number of iterations when compared to the conventional methods. Also, the standard deviation in the estimated parameters upon repeated application of the EHO was about 52% lower than the conventional methods on an average. The implications of inaccurate predictions of the transport parameters from conventional algorithms on the adsorption kinetics are discussed.