Volcanic ash is one of the major potential hazards from volcanic eruptions. It can have both short-range effects from proximal ashfall and long range impacts from volcanic ash clouds. The timely tracking and understanding of recently emitted volcanic ash clouds is important, because they can cause severe damage to jet aircraft engines and shut down major airports. Dispersion models play an important role in forecasting the movement of volcanic ash clouds by being the only means to predict a clouds' trajectory. Where available, comparisons are possible to both remote-sensing data and observations from the ground and aircraft. This was demonstrated in January 2006, when Augustine Volcano erupted after about a 20-year hiatus. From January 11 to 28, 2006, there were 13 explosive events, with some lasting as long as 11 minutes and producing ash clouds as high as 10-12 km (33,000-39,000 ft) above mean sea level (a.m.s.l). From January 28 to February 4, 2006, there was a more continuous phase, with ash clouds reaching 4-5 km a.m.s.l (13,000-16,000 ft). During the eruption, the Puff dispersion model was used by the Alaska Volcano Observatory for trajectory forecasting of the associated volcanic ash eruption clouds. The six explosive events on January 13 and 14, 2006, were the first time the 'multiple eruptions' capability of the Puff model was used during an eruption response. Here we show the Puff model predictions made during the 2006 Augustine eruption and compare these predictions to satellite remote-sensing data, Next Generation Radar (NEXRAD) radar, and ashfall measurements. In addition, we discuss how automated predictions for volcanoes at elevated alert status provide a quicker assessment of the risk from the potential ash clouds.