Theory, modeling, and observations indicate that marine ice sheets on a retrograde bed, including Thwaites Glacier, Antarctica, are only conditionally stable. Previous modeling studies have shown that rapid, unstable retreat can occur when steady ice-shelf basal melting causes the grounding line to retreat past restraining bedrock bumps. Here we explore the initiation and evolution of unstable retreat of Thwaites Glacier when the ice-shelf basal melt forcing includes temporal variability mimicking realistic climate variability. We use the three-dimensional, higher-order Model for Prediction Across Scales-Land Ice (MPASLI) model forced with an ice shelf basal melt parameterization derived from previous coupled ice sheet/ocean simulations. We add sinusoidal temporal variability to the melt parameterization that represents shoaling and deepening of Circumpolar Deep Water. We perform an ensemble of 250 year duration simulations with different values for the amplitude, period, and phase of the variability. Preliminary results suggest that, overall, variability leads to slower grounding line retreat and less mass loss than steady simulations. Short period (2 yr) variability leads to similar results as steady forcing, whereas decadal variability can result in up to one-third less mass loss. Differences in phase lead to a large range in mass loss/grounding line retreat, but it is always less than the steady forcing. The timing of ungrounding from each restraining bedrock bump, which is strongly affected by the melt variability, is the rate limiting factor, and variability-driven delays in ungrounding at each bump accumulate. Grounding line retreat in the regions between bedrock bumps is relatively unaffected by ice shelf melt variability. While the results are sensitive to the form of the melt parameterization and its variability, we conclude that decadal period ice shelf melt variability could potentially delay marine ice sheet instability by up to many decades. However, it does not alter the eventual mass loss and sea level rise at centennial scales. The potential differences are significant enough to highlight the need for further observations to constrain the amplitude and period of the modes of climate and ocean variability relevant to Antarctic ice shelf melting.