Watershed management (WM) aims at enhancing the water availability in rainfed areas through water conservation structures, which facilitate storage of water and recharge to ground water. Identification of suitable locations for placing these structures play a major role in the effectiveness of the water conservation. Site suitability evaluation of water conservation structures is performed through an assessment of various biophysical and socio-economic factors. Many of these factors are expressed in linguistic terms rather than precise numeric values, and therefore the output of the evaluation gets subjective. In this study, a fuzzy inference system (FIS) is developed for site selection of water harvesting structures (check dams, farm ponds, and percolation tanks), owing to its capability to handle linguistic data effectively. The suitability zones were identified using the slope, soil permeability and runoff potential as input variables to the FIS. Trapezoidal membership function (MF) was considered for the input and output variables for the fuzzy model and MF parameters were obtained from literature and expert knowledge. The developed FIS is illustrated through an application to Kondepi watershed, Andhra Pradesh, India. The FIS categorized the majority of the watershed area into high suitability class for both farm ponds and check dams. However, the watershed characteristics were not conducive for percolation tanks according to the FIS. A sensitivity analysis of the FIS parameters suggested that the check dam suitability was sensitive to the soil permeability classes. The suitability maps from the FIS were in good agreement with the location of the existing structures in the watershed, suggesting potential use of the developed FIS in WM decisions. © 2019 Elsevier B.V.