Knowing the spatial and temporal variation of rainfall is crucial for the management of agricultural productivity. This study aimed at evaluating rainfall variability impact on crop yields and its relationship with sea surface temperature using GIS and statistical techniques in North and South Wello zones, northern Ethiopian Highlands. In this study, 30 years (1984–2013) time series data of rainfall from National Centers for Environmental Prediction’s Climate Forecast System Reanalysis and National Meteorological Agency, sea surface temperature from National Oceanic and Atmospheric Administration, and crop data from Central Statistical Agency were used. With regard to spatial and temporal distribution of rainfall, the highest amount of mean total Kiremt/summer and annual rainfall was 993.71 mm and 1715.95 mm, respectively. The magnitude of mean total Kiremt and annual rainfall temporal and spatial variations, using percentage of coefficient variation, was ranging from 24.8 to 62% and from 23.13 to 57.86%, respectively. The correlation between total Kiremt and Belg rainfall with mean sea surface temperature was negative (r = −0.457) and positive (r = 0.385), respectively and statically significant at alpha 0.05 level. From the regression analysis, it was observed that the variation of crop production was explained by the variation of rainfall. For instance, the impact of mean Kiremt rainfall accounted for 47.5% (R2 = 0.475) of variations in sorghum. Generally, the analysis of this study revealed that rainfall variability is one of the major determinant factors for variation of major crop yields in the study area, and this information might be used as an input for decision-makers take appropriate adaptive measures in various agricultural and water resource sectors. For instance, this study recommends the expansion of irrigation and water management practice and promoting perennial trees like fruits and drought-resistant crops.