ln Madagascar, rice is cultivated on 1.3Mha of which 29% are upland rice, growing Irom the coastal area up to the higher altitude. ln the mid-1980s, CIRAD and FOFIFA launched a research program lor the highlands with the aim of pushing forward the frontier of upland rice growing areas in high elevation areas. Today, upland rice is a part of the Madagascar Highland?s landscape and creates new breeding challenges. Low temperatures slow down rice grow1h at almost ail stages: panicle initiation is delayed and cold conditions during the reproductive stage may induce high sterility rate. Thermal environment also is known to affect the speed of vegetative development of the crop, thus the crop duration itsel!. Climate change is assumed to result in arise of mean Temperatures of 2-5 degrees depending on the simulation scenario. Thus rice cropping in higher altitudes may become more favorable as long as precipitation is not a limiting factor. Fields experiments were conducted in three locations along an altitudinal gradient, by using ten contrasting upland genotypes with 5 monthly staggered planting dafes. Physiological and Phenological responses, grain yield and yield components, harvest index and sterility were observed in view of detecting genotypic dillerences across changing environments. In ail ten varieties, crop duration was longer in higher altitude as compared to lower altitude while Harvest index was lound to be higher in lower altitude. Low temperature effect was revealed by the rate of sterility. The percentage of filled spikelet was linked to the minimal Temperature between booting and heading stage and a linear relationship was used to detect the threshold Temperature leading to sterility for ail varieties studied. CoId tolerant varieties adapted to high altitude showed higher yield in high altitude when sown in the recommended sowing date and perlormed weil in both favorable and unfavorable environment. These varieties can respond favorably in high altitude u