Giant subduction megathrust earthquakes of magnitude 9 and larger pose a significant tsunami hazard in coastal regions. In order to test and improve empirical tsunami forecast models and to explore the susceptibility of different subduction settings we here analyze the scaling of subduction earthquake-triggered tsunamis in the near field and their variability related to source heterogeneities. We base our analysis on a sequence of 50 experimentally simulated great to giant (Mw = 8.3–9.4) subduction megathrust earthquakes generated using an elastoplastic analog model. Experimentally observed surface deformation is translated to local tsunami runup using linear wave theory. We find that the intrinsic scaling of local tsunami runup is characterized by a linear relationship to peak earthquake slip, an exponential relationship to moment magnitude, and an inverse power law relationship to fore-arc slope. Tsunami variability is controlled by coseismic slip heterogeneity and strain localization within the fore-arc wedge and is characterized by a coefficient of variation Cv ∼ 0.5. Wave breaking modifies the scaling behavior of tsunamis triggered by the largest (Mw > 8.5) events in subduction settings with shallow dipping (