Abstract.  1. The effects of climate change on species’ ranges have been usually inferred using niche‐based models creating bioclimatic envelopes that are projected into geographical space. Here, we apply an ensemble forecasting approach for niche models in the Neotropical grasshopper Tropidacris cristata (Acridoidea: Romaleidae). A novel protocol was used to partition and map the variation in modelled ranges due to niche models, Atmosphere‐Ocean Global Circulation Models (AOGCM), and emission scenarios.2. We used 112 records of T. cristata and four climatic variables to model the species’ niche using five niche models, four AOGCMs and two emission scenarios. Combinations of these effects (50 cross‐validations for each of the 15 subsets of the environmental variables) were used to estimate and map the occurrence frequencies (EOF) across all analyses. A three‐way anova was used to partition and map the sources of variation.3. The projections for 2080 show that the range edges of the species are likely to remain approximately constant, but shifts in maximum EOF are forecasted. Suitable climatic conditions tend to disappear from central areas of Amazon, although this depends on the AOGCM and the niche model. Most of the variability around the mapped consensus projections came from using distinct niche models and AOGCMs.4. Although our analyses are restricted to a single species, they provide new conceptual and methodological insights in the application of ensemble forecasting and variance partition approaches to understand the origins of uncertainty in studies assessing species responses to climate change in the tropics.