Agricultural ecosystems (or ‘agroecosystems’), comprising about 40% of the world’s land area, are critical for providing food and livelihoods for people globally. Typically, agroecosystems contain only small and fragmented pieces of ‘non-production’ vegetation, embedded within larger areas of crops or pastoral grazing areas. The types of non-production vegetation encountered in agricultural landscapes might include features such as patches of original, pre-agricultural forest, planted hedgerows, streamside vegetation, and patches of herbaceous plants growing along field margins. In theory, these pieces of non-production vegetation can be extremely important for two main reasons. First, they provide critical habitat for species and therefore contribute to the conservation of biodiversity in these highly-modified landscapes. Second, non-production vegetation can help regulate important processes, such as the flow of water and nutrients, the pollination of crops and other plants, the movement of birds and other animals, and therefore contribute to maintaining a functioning, healthy agroecosystem. Does previous research provide evidence for these roles of non-production vegetation across the globe? What methods do researchers use in these studies, and what types of non-production vegetation and ecosystem processes are most commonly studied? To answer these questions, our research team assessed the information presented in 342 published scientific studies. Our analysis showed that non-production vegetation overwhelmingly enhanced biodiversity and the function of ecosystem processes. However, only 17% of studies directly measured the ecosystem processes of interest; most studies only examined indicators of the process. For example, many studies looked at the presence of birds in vegetation patches in agricultural landscapes and, from this, inferred how the animals are moving around and using these patches. We conclude from our analysis that, firstly, it would be useful for future studies to be designed in a way that could directly test how non-production vegetation affects key ecosystem processes. Secondly, we emphasise that when carrying out such research, it is important to consider how human decision making can affect how non-production vegetation is managed in agroecosystems, and therefore how it can be used to provide the best outcomes for both humans and the ecosystem.