Carbon labelling has been proposed as a strategy to encourage cafeteria diners to reduce meat consumption, choose lower carbon-emitting meals and contribute towards global climate change targets. However, field-experimental evidence for label effectiveness is largely limited to trials in universities with student samples. This study evaluated whether the beneficial effects of carbon labelling observed in universities could be replicated in the wider workplace population through a natural field experiment in four call-centre cafeterias in Northern England. Baseline vegetarian uptake in the call-centre cafeterias was significantly lower than in previous university trials (7% vs 15–66%), potentially reflecting a more meat-attached population. The introduction of labels resulted in a 1.5 percentage-point shift from meat to vegetarian meals compared to 1.7–4.6 percentage-point shifts observed in university trials. The increase was almost entirely driven by higher vegetarian sales during the first week of the intervention, with little to no effect observed in subsequent weeks. No statistically significant changes were found in average cafeteria emissions. The findings suggest that carbon labels are not a panacea, and where worksite cafeterias have ambitious emissions targets, labels will need to be implemented alongside other measures.