The crop-livestock integrated farming system practiced in most developing countries depends to a greater extent on the ecosystem as a whole. The importance of animals as an agent of nutrient recycle, sources of rural energy in terms of draft power and fuel as well as major contributor of the farm economy, has resulted into increased population of ruminant stock in these regions creating threats to the sustainability and productivity of land resources. This case study research attempted to formulate optimum herd size compatible to different resource holding farm categories within the sub watershed in mid hills region of Nepal. The research was conducted by classified data collection in Nepal and analysis using Linear Programming (LP) techniques. The LP analysis revealed that the farmers of large, medium and small categories of farms can optimize their livestock holding with combination of 3 Livestock Units (LU) buffaloes and 4 LU goats, 2 LU buffaloes and 4 LU goats and 1 LU buffaloes and 4.4 LU goats with maximum return to the farm family without exerting pressure on the fragile natural resources.