Estimates of the zero emissions commitment (ZEC) rely on multi-model ensembles that sample structural differences between models but cannot assign probabilities to outcomes. Here we present, to our knowledge, the first observationally constrained perturbed parameter ensemble (PPE) assessment of ZEC with a process-based Earth system model, using ~340 members of the University of Victoria Earth System Climate Model across ten emission-rate and carbon-budget combinations following and extending the flat10MIP protocol. The resulting distributions of ZEC, 50 years after emissions cessation, are approximately bell-shaped, with median values between –0.1 and 0.16 °C, and interquartile ranges of 0.06–0.15 °C. We find the parametric uncertainty in this single-model study to be of the same order as, though smaller than, the model structural uncertainty assessed in multi-model intercomparisons, and it therefore should be explicitly accounted for in ZEC50 assessments. Furthermore, we find that the ensemble median differs from the published tuned model configuration, suggesting that single ‘best-tuned’ models may not represent the central estimate of the observationally consistent range. Our PPE indicates that cumulative emissions budget controls the median and the probability of post-cessation warming (11%–94%), while emission rate modulates the spread of outcomes. No individual parameter was found that dominates ZEC.