Reducing the prevalence of fat-rich, energy-dense diets is a public health priority. The present parallel-designed randomised study compared three interventions aimed to increase intakes of low-fat starchy foods and to reduce fat intakes among 589 individuals from 169 families in the Family Food and Health Project (FFHP). Intervention A was education only, intervention B provided ‘cook and eat’ sessions only, whereas intervention C included personalised goal setting, ‘cook and eat’ and education. Diet was assessed at baseline (T0) and at 3 months (T1), 6 months (T2) and 18 months (T3) post-intervention. Retention rates were 75 % at T1, 63 % at T2 and 40 % at T3. ANCOVA (baseline intake as covariate) was assessed between intervention differences at T1, T2 and T3. At T1, individuals in intervention C consumed less fat (P = 0·02) and more total carbohydrate (P = 0·001), starch (P = 0·04) vitamin C (P = 0·002) and NSP (P = 0·01) than those in intervention A. Whereas similar dietary intakes were reported across interventions at T2, participants in intervention C had less energy-dense diets that contained more NSP and vitamin C at T3 than intervention A (P < 0·0001, P = 0·002 and P = 0·01, respectively). Across all intervention groups, the more socially deprived participants in the FFHP (n 119) consumed less fat (P = 0·01) and more total carbohydrate (P = 0·02) at T2 than the least socially deprived (n 240). These data demonstrate the importance of personalised goal setting to translate knowledge and practical cooking skills into healthier food choices, suggesting that low-fat starchy food-focused interventions may be effective in reducing fat intake.