In Brazil, most malaria notifications in the national surveillance database SIVEP are due to Plasmodium vivax. We apply a spatio-temporal model to study the trends of malaria cases in the Brazilian Amazon from 2003 to 2018. We also study possible factors, such as deforestation, human migration and changes in population occupations to understand recent increases occurring after 2015. We found multifactorial drivers for the rebound of malaria cases, pointing out which states had these factors over time significantly associated with the time series of malaria cases. In states Amazonas and Pará, an increase by 1% in cases with people in outdoor occupations led to an estimated values of 8.2% and 3.8% of increase overall, respectively. In Pará and Rondônia states, increases of 1% in malaria cases in agriculture work led to 1.3% of increases in overall P. vivax cases. Recent human migration might have an effect in number of cases, however most migration was in the North of the Brazilian Amazon, far from the border of Rondônia, where a significant downward change was perceived. These results point to evidence that the work on malaria elimination requires efforts on multiple fronts.