Earth system model ensembles exhibit considerable uncertainties surrounding trends and magnitude of steric sea-level variations, due in part to structural model differences, internal model variability, and parameterizations that influence ocean heat uptake. Here we analyze depth-integrated steric sea-level changes using the CMIP5 models and a new CESM ensemble that samples internal variability of the coupled Earth system. The CESM ensemble contains 50 members, with historical and future projections (1850-2100) initialized from unique model states sampled from a ~10,000 year fully coupled unforced equilibrium simulation. The CESM ensemble enables us to examine how initial conditions uncertainty (internal variability) within the full-ocean can influence depth-integrated steric sea-level variability. The second ensemble is comprised of runs from 32 different CMIP5 models. We performed grid-level drift correction for each model using the pre-industrial control simulations, which enables us to examine depth-integrated variability and trends due to different model structures. We compare and contrast our results with published observational datasets, and we analyze the effect of different sources of uncertainty on simulated sea-level variability and trends for different ocean depths. Results point to the importance of the deep ocean in attempting to attribute and predict temporal patterns of steric sea-level on a global scale.