Among the events of the 2009 eruption at Redoubt Volcano, Alaska, event 5 was the best documented by radar, satellite imagery, and deposit mapping. We use the new Eulerian tephra transport model Ash3d to simulate transport and deposition of event 5 tephra at distances up to 350 km. The eruption, which started at about 1230 UTC on 23 March, 2009, sent a plume from the vent elevation (estimated at 2.3 ± 0.1 km above sea level or a.s.l.) to about 16 ± 2 km above sea level in 5 min. The plume was a few kilometers higher than would be expected for the estimated average mass eruption rate and atmospheric conditions, possibly due to release of most of the eruptive mass in the first half of the 20-minute event. The eruption injected tephra into a wind field of high shear, with weak easterly winds below ~ 3 km elevation, strong southerly winds at 6–10 km and weak westerlies above ~ 16 km. Model simulations in this wind field predicted development of a northward-migrating inverted “v”-shaped cloud with a southwest-trending arm at a few kilometers elevation, which was not visible in IR satellite images due to cloud cover, and a southeast-trending arm at > 10 km elevation that was clearly visible. Simulations also predicted a deposit distribution that strongly depended on plume height: a plume height below 15 km predicted ash deposits that were located west of those mapped, whereas good agreement was reached with a modeled plume height of 15–18 km. Field sampling of the deposit found it to contain abundant tephra aggregates, which accelerated the removal of tephra from the atmosphere. We were able to reasonably approximate the effect of aggregation on the deposit mass distribution by two methods: (1) adjusting the grain-size distribution, taking the erupted mass < = 0.063 mm in diameter and distributing it evenly into bins of coarser size; and (2) moving 80–90% of the mass < = 0.063 mm into a single particle bin ranging in size from 0.25 to 1 mm. These methods produced an area inside the 100 g m^− 2 isomass lines that was within a few tens of percent of mapped area; however they under-predicted deposit mass at very proximal (< 50 km) and very distal (> 250 km) locations. Modeled grain-size distributions at sample locations are also generally coarser than observed. The mismatch may result from a combination of limitations in field sampling, approximations inherent in the model, errors in the numerical wind field, and aggregation of particles larger than 0.063 mm.