A pilot trial was conducted to test adherence to specific lifestyle interventions among Pima Indians of Arizona, and to compare them for changes in risk factors for diabetes mellitus. Ninety-five obese, normoglycaemic men and women, aged 25-54 years, were randomized to treatments named 'Pima Action' (Action) and 'Pima Pride' (Pride), which were tested for 12 months. Action involved structured activity and nutrition interventions, and Pride included unstructured activities emphasizing Pima history and culture. Adherence to interventions, changes in self-reported activity and diet, and changes in weight, glucose concentrations, and other risk factors were assessed regularly. Thirty-five eligible subjects who had declined randomization were also followed as an 'observational' group and 22 members of this group were examined once at a median of 25 months for changes in weight and glucose concentration. After 12 months of intervention, members of both intervention groups reported increased levels of physical activity (median: Action 7.3 h month(-1), Pride 6.3 h month(-1), p < 0.001 for each), and Pride members reported decreased starch intake (28 g, p = 0.008). Body mass index, systolic and diastolic blood pressures, weight, 2-h glucose and 2-h insulin had all increased in Action members (p < 0.003 for each), and waist circumference had decreased in Pride members (p = 0.05). Action members gained more weight than Pride members (2.5 kg vs 0.8 kg, p = 0.06), and had a greater increase in 2-h glucose than Pride members (1.33 mM vs 0.03 mM, p = 0.007). Members of the observational group gained an average of 1.9 kg year(-1) in weight and had an increase of 0.36 mM year(-1) in 2-h glucose. Sustaining adherence in behavioural interventions over a long term was challenging. Pimas may find a less direct, less structured, and more participative intervention more acceptable than a direct and highly structured approach.