The people inhabiting the mountains of the Central Himalayan region of India are heavily dependent on their immediate natural resources for their survival. However, this resource-poor mountain ecosystem is gradually becoming unable to provide a minimum standard of living to its continually growing population. In this ecosystem, human population is doubling every 27–30 years, against the declining resource base, particularly forests. Forest are disappearing both quantitatively and qualitatively. Against the requirement of 18 ha of forest land to maintain production in 1 ha of cultivated land, the ratio of forests to cultivated land is only 1.33: 1. The present production from grasslands supports 8 units of livestock, against the ideal 2 units, and the gap between the demand and deficit of fodder is more than 5-fold. Loss of vegetative cover is resulting in drying up of water resources, compelling the women to walk longer distances to collect water. This ecological deterioration, apart from human growth and interference, is compounded by mountain specificities such as inaccessibility, fragility, marginality, diversity, niche and adaptability. The specificities manifest in isolation, distance, poor communication, limited mobility, etc., resulting in limited external linkages and replication of external experiences, and slow pace of development. They, therefore, restrict options for economic growth, effecting poverty and affecting the quality of life of the people of the region. Poverty, in this mountain ecosystem cannot be understood and assessed independent of ecological wealth and would better be termed as ecological poverty. The development efforts to be effective in alleviating poverty here, should take into account mountain specificities and incorporate options which have larger human dimensions, such as mechanisms for population control, socio-economic and cultural conditioning, indigenous knowledge systems of the local people and simple technologies that are already in practice or have potential and are based on least external inputs.