The appearance of molecular replicators (molecules that can be copied) was probably a critical step in the origin of life. However, parasitic replicators take over, and would have prevented life from taking off, unless the replicators were compartmentalized in reproducing protocells. Paradoxically, control of protocell reproduction would seem to require evolved replicators. We show here that a simpler population structure, based on cycles of transient compartmentalization (TC) and mixing of RNA replicators, is sufficient to prevent takeover by parasitic mutants. TC tends to select for ensembles of replicators that replicate at a similar rate, including a diversity of parasites that could serve as a source of opportunistic functionality. Thus TC in natural, non-biological compartments could have allowed life to take hold. Main text:The earliest molecular replicators (1, 2) must have been plagued by freeloading parasitic replicators (3-6).For example, when the RNA genome of the Q virus was replicated in vitro using the viral replicase, 83% of the genome was deleted due to selection for RNAs with the fastest replication rate (7). Eventually, reproducing compartments (protocells) must have arisen, taming parasites by spatially limiting their propagation and allowing group selection at the compartment level, preventing functional collapse (5,(8)(9)(10). Indeed, serial fusion-division cycles of water-in-oil emulsion