Explanatory crop growth stimulation models, constructed on the underlying physiological processes such as CO2 assimilation and respiration as influenced by environmental factors, can be used to predict different levels of production classified on the basis of various stresses. Agroecological zoning of different parts of India for yield of rubber was carried out under rainfed conditions through simulation in different environments, without accounting for stresses due to nutrients, pests and diseases but taking into consideration standard loss of trees in the life cycle of the crop due to natural damage. Mean yield during the first 20 years of tapping has been simulated under specific assumptions for 27 locations, spread all over the traditional/non-traditional rubber cultivating regions of India. The commercial yield data of the clone RRIM 600 was used for the validation of the simulated yield. The yield per hectare ranged from 500 kg at Nellore in Andhra Pradesh to 1427 kg at Trivandrum in Kerala. Wide variations have been observed in the simulated yield among locations in non-traditional areas. On the basis of the predicted yield, the non-traditional areas were classified into high, medium and low productivity zones. The Andaman-Minicoy islands and locations in North East India are under the high productivity zone. The medium zone consists of locations in Karnataka, Goa, Maharashtra, Orissa and West Bengal. Nellore, Kakinada and Visakhapatnam are grouped under the low productivity zone. The rainfed rubber yield decreased with increase in latitude in the West Coast and its reverse is observed in the East Coast and North East India.