Informal statistical inference has now been researched at all levels of schooling and initial tertiary study. Work in informal statistical inference is least understood in the early years, where children have had little if any exposure to data handling. A qualitative study in Australia was carried out through a series of teaching experiments with a class of five- to six-year-old children in two phases over six months. The aim of the exploratory study was to understand and support the emergence of informal statistical inference in the early years of schooling. Through activities that initially built on children's experiences with prediction and key characteristics of informal statistical inference, the children's actions were observed in a data-based inquiry involving prediction to identify critical relationships that then supported children in making informal statistical inferences. Implications are discussed.