Redesigning choice environments appears a promising approach to encourage healthier eating and physical activity, but little evidence exists of the feasibility of this approach in real-world settings. The aim of this paper is to portray the implementation and feasibility assessment of a 12-month mixed-methods intervention study, StopDia at Work, targeting the environment of 53 diverse worksites. The intervention was conducted within a type 2 diabetes prevention study, StopDia. We assessed feasibility through the fidelity, facilitators and barriers, and maintenance of implementation, building on implementer interviews (n = 61 informants) and observations of the worksites at six (t1) and twelve months (t2). We analysed quantitative data with Kruskall–Wallis and Mann–Whitney U tests and qualitative data with content analysis. Intervention sites altogether implemented 23 various choice architectural strategies (median 3, range 0–14 strategies/site), employing 21 behaviour change mechanisms. Quantitative analysis found implementation was successful in 66%, imperfect in 25%, and failed in 9% of evaluated cases. These ratings were independent of the ease of implementation of applied strategies and reminders that implementers received. Researchers’ assistance in intervention launch (p = 0.02) and direct contact to intervention sites (p < 0.001) predicted higher fidelity at t1, but not at t2. Qualitative content analysis identified facilitators and barriers related to the organisation, intervention, worksite environment, implementer, and user. Contributors of successful implementation included apt implementers, sufficient implementer training, careful planning, integration into worksite values and activities, and management support. After the study, 49% of the worksites intended to maintain the implementation in some form. Overall, the choice architecture approach seems suitable for workplace health promotion, but a range of practicalities warrant consideration while designing real-world implementation.