The adsorption equilibrium of a wide range of phosphorus species by an aluminium-based water treatment sludge (Al-WTS) was examined in this study. Four kinds of adsorption-isotherm models, namely Langmuir, Freundlich, Temkin and Dubinin-Radushkevich, were used to fit the adsorption equilibrium data. In order to optimise the adsorption-isotherm model, correlation coefficient (R 2 ) and four error functions were employed to facilitate the evaluation of fitting accuracy. Experiments have demonstrated that the Al-WTS may be an excellent raw material to adsorb P in polluted aqueous environment with adsorption ability in the order of KH 2 PO 4 (ortho-P) > Na(PO 3 ) 6 (poly-P) > C 10 H 14 N 5 O 7 P· H 2 O (organic-P). More importantly, this study provides an entire comparison of the four isotherms in describing the P adsorption behaviour. By considering both the standard leastsquare based R 2 and the results of four error functions analysis, this study reveals that the Freundlich isotherm appears to be the best model to fit the experimental equilibrium data. Langmuir and Temkin isotherms are also good models in current experimental conditions while DubininRadushkevich isotherm poorly described the adsorption behaviour. The error analysis in this study provides vital evidence to reflect its role in facilitating the optimisation in adsorption isotherm study.Obviously, R 2 seems inadequate in optimising multi-isotherm models due to its inherent bias resulting from the least-squares linearisation.