Introduction Smoking starts and progresses rapidly during adolescence. Therefore, it is important to prevent youths from smoking. Previous research is mostly conducted on adolescent samples. This innovative study will focus on smoking of children aged 9–11 years old. The aim is to test whether parental smoking-specific communication and parental smoking add to smoking-specific cognitions derived from the Theory of Planned Behaviour (TPB) in predicting smoking onset.

Method A total of 1478 pairs of mothers and children participated. Structural equation models in Mplus were used to examine whether parental smoking-specific communication are important in shaping children's smoking cognitions, which affects smoking onset.

Results Pro-smoking attitude, higher frequency of communication and high parental smoking were associated with smoking onset. No significant association was found for self-efficacy, perceived norm of mother, (best) friends on smoking onset. We also found that a frequency of communication, quality of communication and parental smoking were related to smoking cognitions.

Conclusion At this age, smoking-specific communication and smoking behaviour of parents are associated with smoking cognitions. From the cognitions, only attitude is related to smoking onset. The extension of TPB is comparable with earlier research on adolescents. A difference between adolescent and child research is that cognitions have no association with smoking onset. A possible explanation can be that smoking-specific cognitions are not developed at this age because children are not engaged in smoking yet. Results can be used to develop effective family-based smoking prevention programs.