Agriculture is threatened by climate change and environmental degradation while also contributing to these problems. The sector thus plays a key role in both mitigation and adaptation through various on-farm measures, making farmers central to a successful sustainable agricultural transition. A vast body of literature has developed to predict farmers’ pro-environmental attitudes, intentions, and behaviour to develop effective policy for supporting farmers’ adoption of sustainable practices. This literature primarily relies on measures of intentions, self-reported behaviours, and cross-sectional approaches, and a recent meta-analysis posits intentions to adopt as one of the strongest determinants of farmers’ uptake of sustainable agricultural practices. However, across the social and psychological literatures findings report a prevalent intention-action gap, that is a lack of a direct relationship between peoples’ intentions and their actions. This is also observed in farmers’ decision-making, where actual adoption rates of sustainable practices remain low. Moreover, the reasons behind this gap remain largely underexplored. In this paper we leverage government and industry connections to conduct a survey with a representative sample of Norwegian horticultural farmers to examine their intentions to adopt and actual adoption of cover crops, a sustainable farming technique that can offer both mitigation and adaptation benefits. We survey farmers before the 2023 growing season to measure their farming preferences, environmental attitudes and concerns, and their intentions to apply for a cover crops subsidy (i.e. their intention to adopt). We subsequently access records of farmers’ application for the cover crops subsidy at the end of the farming season, eight months later, to measure their actual adoption of cover crops. In line with the intention-action gap we found that whereas 49% intended to apply for the cover crops subsidy, there was a low rate of applications: only 16% of our sample applied for the subsidy in 2023. Overall, intentions to adopt were not a significant predictor of actual adoption in our sample. The predictors of intentions, which are also frequently cited in the literature as predictors, did not predict actual behaviour. Moreover, while the linear relation between intentions and behaviours was not significant, we found a non-linear effect: only the highest level of reported intentions had some value in predicting actual behaviour. We contend that these findings hold important implications for policy. Our results offer timely evidence of the intention-action gap in agricultural decision-making, the urgent need to address a widespread reliance on intentions as predictors of behaviour, and the importance of longitudinal approaches to understand, and possibly overcome, the intention-action gap.