The role of probability in curricula for children has fluctuated greatly over the past several decades. Recently, some countries have removed probability from their preschool and primary curricula, and others have retained it. One reason for such lack of agreement is that theory about early probability learning is still relatively new and under development. The purpose of this report is to sketch a tentative theoretical structure with the potential to anchor curricular decisions and inform further research on early probability learning. Toward that end, we begin by reviewing existing literature. We attend, in particular, to the probabilistic thinking tendencies exhibited by children whom researchers consider to be in the earliest stages of learning the subject. We then use these tendencies to posit two different cycles for early probabilistic thinking. One of these cycles is compatible with disciplinary norms and supports normative thinking; thinking tendencies in this normative compatible cycle include attending to the position of objects in a container, forming images of random generators, attending to the operation of random generators, and thinking about past experiences playing games of chance. Thinking tendencies in the other posited cycle lead to belief systems that conflict with normative disciplinary practice; these belief systems include elements such as myths, superstitions, animism, and determinism. We illustrate the two cycles using empirical data from design-based research. We then reflect on how the two cycles comprising our structure of early probabilistic thinking (SEPT) framework can provide a basis for further curricular and theoretical work.