AbstractPredicting the distribution of oil, buoyant plastics, flotsam, and marine organisms near the ocean surface remains a fundamental problem of practical importance. This manuscript synthesizes progress in this area during the time of the Gulf of Mexico Research Initiative (GoMRI; 2012–2019), with an emphasis on the accumulation of floating material into highly concentrated streaks on horizontal scales of meters to 10's of kilometers. Prior to the GoMRI period, two new paradigms emerged: the importance of submesoscale frontal dynamics on the larger scales and of surface-wave-driven Langmuir turbulence on the smaller scales, with a broad transition occurring near 100 m. Rapid progress resulted from the combination of high resolution numerical modeling tools, mostly developed before GoMRI, and new observational techniques developed during GoMRI. Massive deployments of inexpensive and biodegradable satellite-tracked surface drifters combined with aerial tracking of oil surrogates (drift cards) enabled simultaneous observations of surface ocean velocities and dispersion over scales of 10 m to 10's of kilometers. Surface current maps produced by ship-mounted radar and aerial optical remote sensing systems, combined with traditional oceanographic tools, enabled a set of coordinated measurement programs that supported and expanded the new paradigms. Submesoscale fronts caused floating material to both accumulate at fronts and to disperse as they evolved, leading to higher local concentrations, but increased overall dispersion. Analyses confirmed the distinct submesoscale dynamics of this process and the complexity of the resulting fields. Existing tools could be developed into predictive models of submesoscale statistics, but prediction of individual submesoscale features will likely remain limited by data. Away from fronts, measured rates of accumulation of material in and beneath surface windrows was found to be consistent with Langmuir turbulence, but highly dependent on the rise rate of the material and thus, for oil, on the droplet size. Models of this process were developed and tested and could be further developed into predictive tools. Both the submesoscale and Langmuir processes are sensitive to coupling with surface waves and air-sea flux processes. This sensitivity is a promising area for future studies.