Vertical redistribution of HNO3 through large HNO3-containing particles associated with polar stratospheric clouds (PSCs) plays an important role in the chemistry of the Arctic winter stratosphere. During the RECONCILE (Reconciliation of essential process parameters for an enhanced predictability of Arctic stratospheric ozone loss and its climate interactions) campaign, apparently very large NAT (nitric acid trihydrate) particles were observed by the airborne in situ probe FSSP-100 (Molleker et al., 2014). Our analysis shows that the FSSP-100 observations associated with the flight on 25 January 2010 cannot easily be explained assuming compact spherical NAT particles due to much too short growing time at temperatures below the existence temperature of NAT (TNAT). State-of-the-art simulations using CLaMS (Chemical Lagrangian Model of the Stratosphere; Grooß et al., 2014) suggest considerably smaller particles. We consider the hypothesis that the simulation reproduces the NAT particle masses in a realistic way, but that real NAT particles may have larger apparent sizes compared to compact spherical particles, e.g. due to non-compact m ... mehrorphology or aspheric shape. Our study focuses on the consequence that such particles would have reduced settling velocities compared to compact spheres, altering the vertical redistribution of HNO3. Utilising CLaMS simulations, we investigate the impact of reduced settling velocities of NAT particles on vertical HNO3 redistribution and compare the results with observations of gas-phase HNO3 by the airborne Fourier transform spectrometer MIPAS-STR associated with two RECONCILE flights. The MIPAS-STR observations confirm conditions consistent with denitrification by NAT particles for the flight on 25 January 2010 and show good agreement with the simulations within the limitations of the comparison. Best agreement is found if settling velocities between 100 and 50% relative to compact spherical particles are considered (slight preference for the 70% scenario). In contrast, relative settling velocities of 30% result in too weak vertical HNO3 redistribution. Sensitivity simulations considering temperature biases of ±1 K and multiplying the simulated nucleation rates by factors of 0.5 and 2.0 affect the comparisons to a similar extent, but result in no effective improvement compared to the reference scenario. Our results show that an accurate knowledge of the settling velocities of NAT particles is important for quantitative simulations of vertical HNO3 redistribution.