Crop production relies on the use of natural resources and is a source of greenhouse gas (GHG) emissions. The present study uses survey data from 250 Polish farms to investigate the eco-efficiency of three main crops: winter wheat, winter triticale, and winter oilseed rape. First, the slack-based Data Envelopment Analysis (SBM-DEA) model with undesirable output (GHG emissions) was applied. In the next step, the Generalized Additive Model for Location, Scale and Shape (GAMLSS) was used to explain the efficiency scores. The calculated GHG emissions per hectare of crop were 1.9 tCO2e, 3.2 tCO2e, and 4.3 tCO2e for winter triticale, wheat, and oilseed rape, respectively. Fully efficient farms used significantly less fertilizer (13.6–29.3%) and fuel (16.6–25.3%) while achieving higher yields (14.4–23.2%) and lower GHG emissions per hectare (10.8–17.7%). In practice, this means that efficient farms had a 20–32% lower carbon footprint per kilogram of yield than inefficient farms, depending on the crop. It was also shown that increasing the size of the cultivated area contributed to improving efficiency scores, while no conclusive evidence was found for an influence of economic size or farm type on their performance. Weather conditions had a significant impact on the efficiency score. In general, higher temperatures and precipitation in spring had a positive effect on efficiency, while an opposite relationship was observed in summer.