The recent acceleration and retreat of Humboldt Glacier (19 cm sea level equivalent) raise concerns that mass loss due to ice dynamics could dramatically increase in the future. We estimate the possible range of sea-level rise from the retreat of Humboldt Glacier using the higher-order ice sheet model MALI. We first invert for basal traction and steady state ice temperature in a PDE-constrained optimization that minimizes modeled error in ice surface velocity relative to observations in the year 2007. We then tune an effective pressure-dependent power law basal friction relationship by imposing observed mean retreat rates from 2007–2017. Power-law exponents in the range of 1/7–1/5 minimize the misfit to observed velocity changes, indicating a semi-plastic bed rheology. Next, we tune a calving parameterization based on the von Mises yield criterion to observed velocity and ice front changes from 2007–2017. Finally, we use these tuned parameter values in an ensemble of simulations from 2007 to 2100 under three climate forcings from CMIP5 (two RCP 8.5 forcings) and CMIP6 (one SSP5 forcing), predicting 6.5–9 mm of sea-level rise from Humboldt Glacier. This is a substantial fraction of the 40–140 mm predicted by the ISMIP6 Greenland-wide ensemble and significantly higher than a previous estimate from Humboldt alone (~3.5 mm). Thus, tuning of the basal friction law to observed changes yields a large increase in the projected sea-level rise from Humboldt Glacier. The modeled glacier develops a sizable ice shelf by the end of the century in many model simulations, a result generally not expected for Greenland outlet glaciers. This indicates that future atmospheric temperatures could further accelerate retreat if surface ponding is sufficient to hydrofracture the ice shelf. Our data-constrained study of Humboldt Glacier shows that the modeled evolution of Greenland's marine-terminating outlet glaciers is sensitive to choices in bed rheology and calving physics. Commonly used simplifications to these choices at the ice-sheet scale could lead to significant underestimation of future sea-level rise contribution.