This trial assessed whether a simple clinical tool can be used to stratify patients with diabetes, according to risk of developing foot ulceration. This was a prospective, observational follow-up study of 3526 patients with diabetes (91% type 2 diabetes) attending for routine diabetes care. Mean age was 64.7 (range 15-101) years and duration of diabetes was 8.8 (+/-1.5 SD) years. Patients were categorised into 'low' (64%), 'moderate' (23%) or 'high' (13%) risk of developing foot ulcers by trained staff using five clinical criteria during routine patient care. During follow-up (1.7 years), 166 (4.7%) patients developed an ulcer. Foot ulceration was 83 times more common in high risk and six times more in moderate risk, compared with low-risk patients. The negative predictive value of a 'low-risk score' was 99.6% (99.5-99.7%; 95% confidence interval). This clinical tool accurately predicted foot ulceration in routine practice and could be used direct scarce podiatry resources towards those at greatest need.