To evaluate the efficacy of adding cognitive behavioural treatment (CBT) to either a low-carbohydrate (LC) diet or a low-fat (LF) diet in the treatment of weight loss of obese women, a randomised clinical intervention study was performed. A total of 105 healthy non-pregnant obese women (average age and BMI of 45·4 (sd 10·4) years and 36 (sd 4·3) kg/m2) were randomly allocated to the CBT or control (C) groups; within each group, women were randomly selected to receive either the LC or LF diet during 6 months. The pre-planned primary trial end-point was the weight loss. Differences between the groups were assessed using one-way ANOVA. There were three women (2·8 %) who dropped out, all of them in the CBT group. No differences in the anthropometric and laboratory characteristics at baseline were noted between women in the CBT (n 52) and control groups (n 50). Intention-to-treat analysis showed that weight loss in the CBT-LC (90 (sd 12·3) to 82·1 (sd 12·1) kg) and C-LC (89·4 (sd 10·0) to 85·8 (sd 9·8) kg) groups reached 8·7 and 4·0 %, respectively (P < 0·0001), and in the CBT-LF (87·9 (sd 11·4) to 79·4 (sd 11·8) kg) and C-LF (88·8 (sd 14·5) to 85·3 (sd 14·3) kg) groups it was 9·7 and 3·9 %, respectively (P < 0·05). Weight loss was higher in the CBT-LF group than in the CBT-LC groups (P = 0·049). The present results showed that adding CBT to either the LF or LC diet produced significantly greater short-term weight loss in obese women compared with diet alone. These finding support the efficacy of CBT in breaking previous dietary patterns and in developing healthier attitudes that reinforce a healthier lifestyle.