Abstract Background Young children’s diets are currently suboptimal. Given that mothers have a critical influence on children’ diets, they are typically a target of interventions to improve early childhood nutrition. Understanding the maternal factors which mediate an intervention’s effect on young children’s diets is important, but has not been well investigated. This research aimed to test whether maternal feeding knowledge, maternal feeding practices, maternal self-efficacy, and maternal dietary intakes acted as mediators of the effect of an intervention to improve child diet quality. Methods The Melbourne Infant Feeding Activity and Nutrition Trial (InFANT) Program was a cluster-randomized controlled trial, conducted from 2008–2010. This novel, low-dose, health promotion intervention was delivered quarterly over 15 months and involved educational activities, promotion of peer discussion, a DVD and written materials. Post-intervention, when children were approximately 18 months of age, child diets were assessed using multiple 24-hour recalls and a purpose-developed index of diet quality, the Obesity Protective Dietary Index. Maternal mediators were assessed using a combination of previously validated and purpose-deigned tools. Mediation analysis was conducted using the test of joint significance and difference of coefficients methods. Results Across 62 parents’ groups in Melbourne, Australia, 542 parents were recruited. Post- intervention, higher maternal feeding knowledge and lower use of foods as rewards was found to mediate the direct intervention effect on child diet quality. While other aspects of maternal feeding practices, self-efficacy and dietary intakes did not act as mediators, they were associated with child diet quality. Conclusions Mediation analysis of this novel health promotion intervention showed the importance of maternal feeding knowledge and use of foods as rewards in impacting child diet quality. The other maternal factors assessed were appropriate targets but further research on how to impact these in an intervention is important. This evidence of intervention efficacy and mediation provides important insights for planning future interventions. Trial registration Current Controlled Trials ISRCTN81847050 , registered 23 November 2007.