Long-term zonal demand forecasting is a complex problem for electric distribution systems, particularly when the planner has limited data, and therefore, needs smart approaches along with proper estimations. This paper presents a novel methodology to determine saturation curves (S-curves) of demand zones which are classified based on municipality development plans. The methodology which is based on combination of top-to-bottom and bottom-to-top (also called end-use) approaches, is applied to the network of Akdeniz Distribution Company (DISCO) of Turkey. Genetic algorithm (GA) technique is utilized to determine zonal S-curves by utilizing top-to-bottom projections and expected saturation demands of the zones. Sensitivity analysis shows that the proposed method gives reliable results as long as top-to-bottom results represent the total demand of the zones satisfactorily and the planner defines the constraints reasonably.