The long-lasting effects of post-natal depression (PND) on parenting stress were assessed. It was also asked whether change in parenting stress followed conventional cognitive-behavioural therapy (CBT) treatment for PND and the benefits of targeted parent–infant intervention was assessed. Data were from three sources (1) a longitudinal cohort of post-partum depressed and non-depressed women; (2) a treatment study of 162 depressed women involving CBT, routine care and a comparison group of 162 non-depressed women; (3) a pilot study of a specialized parent–infant intervention with 22 depressed women. The longitudinal cohort showed elevated parenting stress persisting until 3.5 years post-partum. In the treatment study, 73 per cent of depressed women had parenting stress scores indicating clinically dysfunctional mother–infant relations before treatment (compared to less than 3 per cent in the non-depressed group). This rate was still 56 per cent after 12 weeks of CBT, with average parenting stress similar to women in routine care. During 3 weeks of a specialized parent–infant intervention, there was a more rapid decline in parenting stress. Thus, parent–infant difficulties due to PND were persistent and CBT targeted only at maternal mood had limited impact on these. Parenting stress appeared to improve to more normal levels after the specialized mother–infant module and this needs to be confirmed in a fully controlled study. Broadened criteria for identifying ‘best-practice’ treatments for PND would take account of impact on mother–infant interactions. Copyright © 2006 John Wiley & Sons, Ltd.