How predictable is evolution? If we placed similar starting populations into replicated similar environments and then watched evolution happen, would each replicate population evolve in a predictable, deterministic way? Or, is evolution fundamentally unpredictable, dominated by chance events during and after colonization? Repeated evolution, often called ‘parallel evolution’, is taken as strong evidence for the role of natural selection in driving deterministic evolutionary change. I will first discuss our research identifying the factors that contribute to repeated evolution at both the phenotypic and genetic level in threespine sticklebacks, a classic system for the study of parallel evolution. One limitation of our previous studies is that we take a retrospective approach and therefore lack crucial historical data on the founding ancestral populations and ecosystem changes over time, which are contingencies that can shape subsequent evolutionary trajectories. An alternative approach is to conduct forward-in-time experimental evolution studies in which the properties of the founding populations are known. Then, we could ask whether replicate populations follow deterministic eco-evolutionary trajectories, or whether contingencies during and after colonization predominate these trajectories? I will describe just such an experiment in which we are following eco-evolutionary trajectories in real time. We have stocked 9 natural lakes in Alaska with ~10,000 stickleback fish. We will quantify the extent to which these replicate populations evolve in parallel, with respect to genotypes, phenotypes, and ecosystem properties, and compare these trajectories to those predicted by theory. This is the first forward-in-time, highly replicated and whole-ecosystem study of parallel evolution in nature, and it will yield novel and fundamental insights on the roles of historical contingencies and deterministic selective forces in shaping eco-evolutionary trajectories.