Tokamak plasma start-up is a highly dynamic phase in which the inductive toroidal loop voltage must ionise the prefill gas, heat the plasma, and ramp up the plasma current. In particular, the ionisation of impurities can place a significant strain on the available input power because of the associated radiation losses. Modelling these coupled processes is essential for designing reliable start-up scenarios, particularly for future devices such as ITER where the available electric field and prefill pressure are severely constrained [1]. STREAM [2] (STart-up Runaway Electron Analysis Model) is a 0D fluid code for self-consistent tokamak start-up simulations, built on the DREAM framework [3] and following the burn-through model of DYON [4, 5]. It solves coupled particle and energy balance equations together with plasma-current and vessel circuit equations, while accounting for confinement losses, plasma-wall interaction, and neutral screening. Runaway electrons are modelled as a relativistic population that carries current and contributes to the total plasma current. Applying STREAM to a specific discharge requires many input parameters, including sputtering yields, electromagnetic circuit elements, and early (post-breakdown) plasma conditions, many of which are poorly constrained experimentally. To determine which inputs most strongly affect the model response, we performed a global, variance-based sensitivity analysis of STREAM for deuterium ohmic start-up in TCV discharge #84108 with a carbon wall. We define a discrepancy metric from the weighted normalised root-mean-square differences of eight observables relative to a reference simulation: electron temperature, electron density, plasma current, $\mathrm{D}_{\alpha}$, $\mathrm{C}_{\mathrm{II}}$, $\mathrm{C}_{\mathrm{III}}$, $\mathrm{O}_{\mathrm{III}}$, and total radiated power. Using the method developed by Homma and revisited by Saltelli [6, 7], we quantify the influence of 25 uncertain input parameters. For each parameter, we compute a global importance measure, $I_j$, together with quantity-specific importance measures for each observable. Preliminary results suggest a clear ranking of parameter influence, with the initial electron temperature and plasma self-inductance among the main contributors to output variability during burn-through. These findings indicate where improved experimental constraints would be most valuable for the reliable validation of start-up simulations on TCV. ## References 1. P. C. de Vries and Y. Gribov, "ITER Breakdown and Plasma Initiation Revisited", *Nuclear Fusion* **59**(9), 096043 (2019). DOI: 10.1088/1741-4326/ab2ef4. 2. M. Hoppe, I. Ekmark, E. Berger, and T. Fülöp, "Runaway Electron Generation During Tokamak Start-Up", *Journal of Plasma Physics* **88**(3), 905880317 (2022). DOI: 10.1017/S002237782200054X. 3. M. Hoppe, O. Embreus, and T. Fülöp, "DREAM: A Fluid-Kinetic Framework for Tokamak Disruption Runaway Electron Simulations", *Computer Physics Communications* **268**, 108098 (2021). DOI: 10.1016/j.cpc.2021.108098. 4. H.-T. Kim, W. Fundamenski, A. C. C. Sips, and EFDA-JET Contributors, "Enhancement of Plasma Burn-Through Simulation and Validation in JET", *Nuclear Fusion* **52**(10), 103016 (2012). DOI: 10.1088/0029-5515/52/10/103016. 5. H.-T. Kim, A. Mineev, D. Ricci, J.-W. Lee, Y.-S. Na, ITPA-IOS members, and JET contributors, "Benchmarking of Codes for Plasma Burn-Through in Tokamaks", *Nuclear Fusion* **60**(12), 126049 (2020). DOI: 10.1088/1741-4326/abb95c. 6. T. Homma and A. Saltelli, "Use of Sobol's Quasirandom Sequence Generator for Integration of Modified Uncertainty Importance Measure", *Journal of Nuclear Science and Technology* **32**(11), 1164-1173 (1995). DOI: 10.1080/18811248.1995.9731832. 7. A. Saltelli, K. Chan, and E. M. Scott, *Sensitivity Analysis*. Wiley, Hoboken, NJ (2000).