While documenting the vegetation response to climatic changes in mountains, the use of grass phytolith data relies on the ability of phytolith assemblages or indices to differentiate the elevationally stratified vegetation zones. To infer the potential and limitations of grass phytolith assemblages and indices to reconstruct vegetation vis-à-vis climate in the Himalayan mountain regions, we analyzed phytolith assemblages from 66 dominant grasses and 153 surface soils from four different forest types along the c. 130-4000 m a.s.l. elevation gradients in the Darjeeling and Arunachal Himalayas. Grass short cell phytolith assemblages from modern grasses show significant variability with rising elevation. To test the reliability of the above observation, phytoliths from the soil samples were subjected to linear discriminant analysis (DA). DA classified 85.3% and 92.3% of the sites to their correct forest zones in the Darjeeling and Arunachal Himalayas respectively. Relative abundance of bilobate, cross, short saddle, plateau saddle, rondel and trapeziform types allow discrimination of the phytolith assemblage along the elevation gradient. Canonical correspondence analysis (CCA) on the soil phytolith data further revealed their relationships with the climatic variables. Temperature and evapotranspiration were found to be the most influential for differential distribution of grass phytolith assemblages with rising elevation in the eastern Himalayas. We also tested the reliability of phytolith indices (Ic, Iph and Fs) for tracing the dominance of different grass subfamilies in the eastern Himalayas. Ic proved to be most reliable in discriminating C3/C4 grass along the elevation gradient while Iph and Fs proved to be less reliable. We observed that in the monsoon dominated eastern Himalayas, a little adjustment in Ic index may enhance the accuracy of interpretations. In future studies more precise identification of phytolith sub-types from additional sites in the eastern Himalayas may further improve our interpretations related to past climate condition. © 2016 Elsevier Ltd.