Abstract A study is described in which the normal toes of 50 patients with unilateral ingrowing toenails were compared with those of 50 healthy subjects matched for age and sex. Twenty-one potential risk factors were compared between the groups. All measurements on patients were derived from unaffected toes in order to avoid anatomical distortion due to the disease process. Statistical analysis revealed significant differences between the groups for three anatomical measures, namely nail fold width (P < 0·001), medial rotation (eversion) of the great toe (P < 0·02) and nail thickness (P < 0·03). Possible discriminating variables were assessed using linear discriminant function analysis and a mathematical equation was derived based on these three factors. This equation was found to possess a potentially high predictive value in that it correctly classified 86 per cent of the patients and 77 per cent of the controls. A further group of 62 unselected patients presenting with unilateral onychocryptosis and 35 unaffected controls were assessed for these three parameters. Of these, 80 per cent of controls and 85 per cent of patients could be classified correctly by the equation. It is thus proposed that, in the majority of subjects with onychocryptosis, a discrete anatomical predisposition exists and that the other factors act only as triggers in the development of the condition. This study provides a rationale for less traumatic surgical approaches to ingrown nail than nail avulsion, with or without nail bed ablation and may also provide a means of identifying high-risk groups.