In this research, the slump and density of lightweight aggregate concrete, which are made with natural Pumice, were predicted by linear regression analysis and adaptive neuro-fuzzy inference system (ANFIS). Due to the multiplicity of parameters affecting density and slump of lightweight aggregate concrete, their precise measurement are time- consuming and so, estimates of them are valuable. For selecting apt ANFIS network, firstly 100 lightweight concrete mixes designed and constructed in concrete laboratory and then their measured characteristics divided into training and testing subsets. Optimum ANFIS network predictions of slump and density are compared with from linear regression estimates and laboratory measured ones. The results indicate that ANFIS is a potent tool for this purpose.
