Description: Deposits and transport processes resulting from the resedimentation of cold, unconsolidated ignimbrite into water were simulated by flume experiments. The ignimbrite sample used was poorly sorted (σ = 2·4–3), fine ash-rich (〈 63 μm, 17–30 wt%) and included both dense lithic clasts (〉 2000 kg m−3) and pumice (500 to ca 1300 kg m−3). As a result of the binding forces of the ash matrix, the experiments involved resedimentation from a steep front onto the floor (with or without an initial ramp) of the water-filled tank under both still and wave-generated conditions. Larger discrete collapse events were induced by oversteepening the sample front and by undercutting from wave action. The mass of the collapse and proportion of pore–space water strongly influenced the style of resedimentation and the deposits. Initial collapse events were from the top of the steep front and fell onto the floor. The largest, densest clasts were deposited as a lithic lag in a proximal sediment wedge or rolled down to a break-in-slope. Fine ash was transported in dilute turbidity currents, and coarse unsaturated pumice clasts floated off. Moderate collapse events generated high-density turbidity currents, trapping pumice in the flow, causing them to saturate. These low-density pumice clasts were easily remobilized by wave activity and passing currents and accumulated on the gentle slope at the bottom of the resedimented deposit. Large collapse events slumped, producing poorly sorted mounds similar in texture to the original starting material. As the matrix of the ignimbrite sample became saturated with water, moderate and large collapse events generated debrisflows and slurries that deposited massive, poorly sorted deposits. Furthermore, once more gentle slopes were established between the sample and deposit, small cascading grainflows deposited lithic clasts on the upper slopes and levees of pumice at the terminus of low-relief, ash channels. The experiments show that, excluding large collapse events and debrisflows, resedimenting ignimbrite in water is effective at segregating low-density pumice clasts from dense lithic clasts and fine ash. Experiments using fine-ash poor ignimbrite and well-sorted quartz sand for comparison formed an inherently unstable initial steep front that immediately collapsed by continuous grain avalanches. The grainflow deposits had textures similar to the fines-poor starting material.