This paper presents an epistemological analysis of typical didactical activities noted in early-years science lessons, which was carried out in an attempt to diagnose the extent to which the teaching practices adopted by early-years educators are successful in supporting young children's understanding in science. The analysis of didactical activities used a framework that allowed us to discover whether they promoted desired connections between theoretical ideas, evidence and the material world. Theoretical ideas, evidence and the material world are entities internal to scientific inquiry and, in educational contexts, connections between them are considered essential in assisting the development of young children's scientific thinking. The results indicated that in the early-years science classroom scientific activity was mainly confined to the representational level. Intervention practices into the material world were limited, and were based on collected evidence. No interventions based on ideas were identified in the science lessons. Missing links between evidence and theory and between ideas and the material world suggest that the didactical activities analysed did not promote scientific understanding. (Contains 2 tables.)