This paper addresses the development of a fuzzy model for weld quality prediction. Significant underwater friction stir welding parameters affecting the weld quality are rotational speed, traverse speed, and tool shoulder diameter‎. Welding experiment is performed on AA 1050 aluminium pipes by central composite design to attain maximum tensile strength and hardness of the weld joint. Quality of weld measured in terms of tensile strength and hardness is predicted using fuzzy logic and the results are compared with statistical analysis. Confirmatory experimental results show that the fuzzy model can predict an adequate output with less error than statistical analysis.