Aiming to determine the interaction between climatic factors, flowering and yield of dry arabica coffee in municipalities in the regions of Southern Minas Gerais and Alto Paranaíba, both in the State of Minas Gerais, this study considered data from the harvest years 2008/2009 to 2017/2018, and climatic data analyzed according to the coffee phenological cycle. The determination of significant variables used the Stepwise method and the Akaike Information Criterion (AIC). The association of yield with climatic variables used a generalized linear model with gamma distribution and logarithmic function. Average flowering (20% to 50% of flowers per branch) used a logistic regression model. The productive variability of arabica coffee showed an association with the biennial cycle, the occurrence of medium and main flowering, flowering period and rains in bean formation. The occurrence of average flowering and the frequency of rainfall during bean formation enhanced coffee yield, at the same time as the increment in maximum temperature in vegetative bud formation reduced flowering and yield.
