In contemporary ocean science, modeling systems that integrate understanding of complex multiscale phenomena and utilize efficient numerics are paramount. Many of today's fundamental ocean science questions involve multiple scales and multiple dynamics. A new generation of modeling systems would allow to study such questions quantitatively by being less restrictive dynamically and more efficient numerically than more traditional systems.Such multiscale ocean modeling is the theme of this topical collection. Two large international workshops were organized on this theme, one in Cambridge, USA (IMUM2010), and one in Bremerhaven, Germany (IMUM2011). Contributions from the scientific community were encouraged on all aspects of multiscale ocean modeling from ocean science and dynamics to the development of new computational methods and systems. Building on previous meetings (e.g., Deleersnijder and Lermusiaux 2008;Deleersnijder et al. 2010), the workshop discussions and the final contributions to the topical collection are summarized next.The scientific application domains discussed and presented ranged from estuaries to the global ocean, including coastal regions and shelf seas. Multi-resolution modeling of physical, biological, chemical, and sea ice processes as well as air-sea interactions were described. The multiscale dynamics considered involved hydrostatic, non-hydrostatic, turbulent, and sea surface processes.Computational results and discussions emphasized multiresolution simulations using unstructured and structured meshes, aiming to widen the range of resolved scales in space and time. They included finite volume and finite element spatial discretizations, high-order schemes, preconditioners, solver issues, grid generation, adaptive modeling, data assimilation, coupling with atmospheric or biogeochemical models, and distributed computing. The advantages of using unstructured meshes and related approaches in particular multi-grid embedding, nesting systems, wavelets, and other multiscale decompositions were discussed. Techniques for the study of multi-resolution results, visualization, optimization, model evaluations, and uncertainty quantification were also examined.