Repeat expanded RNA molecules aggregate under certain conditions both in vitro and in vivo. Understanding the mechanism for aggregation—including how aggregation properties change with sequence and environmental conditions—would explain and predict the behavior of RNA-based biomolecular condensates, and enable the rational design of RNA-based materials. Here, we introduce an analytical framework to predict aggregation for any repeat RNA sequence, accounting for both intra- and inter-molecular bonding. By enumerating the equilibrium landscape of multimers, we reveal the driving force for aggregation: the increased configurational entropy associated with the multiplicity of ways to form bonds in the aggregate. Our model uncovers rich phase behavior, including a sequence-dependent reentrant phase transition, and repeat parity-dependent aggregation. We validate our results by comparison to a complete computational enumeration of the landscape, and to previously published molecular dynamics simulations. Our work unifies and extends published results, and enables the design of programmable RNA condensates.