We analyzed the spatial patterns (1985-2018) of planted area and productivity of beans, cassava and corn, the main crops grown under rainfed conditions intended for food production in family farming in the Brazilian semiarid region. The spatial construction used data collected from Municipal Agricultural Surveys combined with remote sensing data. In general, the maps indicate heterogeneous patterns, with a marked decrease in the planted area, in the spatio-temporal evolution of agricultural variables. Cassava is the one with the lowest occupation of planted areas (~11%), followed by beans (~43%) and corn (~46%). On the other hand, beans are the crop that lost the most growing space in the study region (37 thousand ha), followed by cassava (14 thousand ha) and corn (12 thousand ha) per year. In relation to the productivity of the crops analyzed, the corn crop stands out, which, even with the reduction of the planted area, increased productivity by more than 160%. Bean and cassava crops reduced their yields by 10% and 34%. The results generated indicate a drop, both in the planted area and in productivity for the three crops analyzed (except corn productivity), which raises awareness of local food security and advances in climate change.