A watershed is a hydrological unit of space composed of interrelated drainage area with a common movement of water including all the consequential implications for land and water use. A micro-watershed is a coherent ecosystem at the smallest viable geographical area. It is administratively as well as operationally the most meaningful planning unit. Micro-watershed planning contributes to sustainable to sustainable development of the area by integrating varied development programmes through the efficient use, equitable access and decentralised control of water resources. This paper presents a planning model for micro-watershed development. Three objectives are considered as adequate to represent the sustainable development ethos. These are economic benefits esentially accruing to land and cattle owners, employment benefits accruing to labouring sections and soil conservation which contributes to long-term sustainability of land. Project options available to a planner are check dams, gully checks, contour bunds, pond development pasture development and afforestation. Constraints include investment budget, lower limit on benefit-cost ration, availability of land for pasture development, afforestation and pond construction. Besides these there are constraints to ensure system relationship logic. The model formulation results in a multiple objective mixed-integer linear formulation wherein objectives and constraints are linear functions of decision variables. A real-life application of the model is presented. A representative group of community is considered as a decision maker (DM). DM's preferences are modelled as a value function which is encoded through a series of questions and answers. We have termed the value function as "Sustainability Index" as it evaluates in a single measure the efficacy of the plan in terms of sustainable development. The decision problem is to select a development plan that maximises Sustainability Index. For the present application, the planning problem results in a concave objective function with eighteen linear constraints, twenty five zero-one and five continuous variables. An algorithm that uses a branch and bound procedure together with an mplicit enumeration logic is developed to solve the problem. The algorithm solves a continuous non-linear problem at each node of the branch and bound tree. Compared to a plan developed by a local agency, the optimal plan improves achievement of each objective by twenty five percent. Sensitivity analysis with varying budget limits is carried out which recommends additional investment beyond the current budget. Improvement of the model through integration of energy and environment options and some features of suitable database to aid the planner are presented.