The El Niño-Southern Oscillation (ENSO) is a major driver of climate and agricultural productivity variability, as those events modulate significant fraction of rainfall and air temperature interannual variation, key factors directly affecting crop performance. This study aimed to evaluate the impacts of ENSO on rice, common bean, and soybean yields in two periods (1961–1990 and 1991–2019) using a Random Forest modeling approach. The analysis focused on the states of Rio Grande do Sul (RS) and Goiás (GO), Brazil, which represent contrasting climatic regions within the country. Daily data on rainfall and maximum and minimum air temperature (1961–2019) were obtained from 69 weather stations. ENSO phases were classified using the Oceanic Niño Index (ONI), in the Niño 3.4 region, and analyzed through Functional Data Analysis (FDA) and Functional Analysis of Variance (FANOVA) across two periods (1961–1990 and 1991–2019). Observed crop yield data were incorporated into Random Forest models to estimate the relative importance of climatic variables (air temperature and rainfall) and spatial factors (municipality) in determining agricultural productivity. The results demonstrate a clear influence of ENSO on climate variability and crop yields, particularly in Rio Grande do Sul, whereas its effects were less pronounced in Goiás. The El Niño phase generally favored yield increases across all evaluated crops (irrigated rice, common bean, and soybean), while the La Niña phase was frequently associated with yield reductions. During Neutral years, the influence of ENSO was comparatively weaker. A reduction in the relative importance of rainfall was observed in the more recent period (1991–2019), indicating increased rainfall homogeneity over time. In contrast, local factors became more influential in determining crop yield. Nevertheless, ENSO remains a critical factor for explaining yield variability and for supporting agricultural management and decision-making strategies.