We discuss fast ejecta observed at 1 AU during a period of increasing solar activity from February 5, 1998, to November 29, 1999. Fast ejecta are transient, noncorotating flows that move past the Earth during a day or more, with a maximum speed >600 km s-1. We identify two classes of fast ejecta at 1 AU: (1) magnetic clouds, whose local magnetic structure is that of a flux rope; and (2) complex ejecta, which are not flux ropes and have disordered magnetic fields. Nearly equal numbers of magnetic clouds and complex ejecta were found: four and five, respectively. The complex ejecta had weaker magnetic fields and higher proton temperatures than the magnetic clouds on average. The average &bgr; for the complex ejecta (0.25¿0.09) was larger than that for the magnetic clouds (0.06¿0.04). The complex ejecta and magnetic clouds had comparable speeds on average, namely, 558¿80 and 500¿63 km s-1, respectively. Using the duration of the stream and that of the counterstreaming electrons to measure the ejecta, the average time for the complex ejecta to move past ACE was 3 days, which is more than twice that for the magnetic clouds. All of the magnetic clouds contain some material with a high &agr;/proton density ratio (>8%) and a density ratio of O7+/O6+>1. However, three of the five complex ejecta did not contain material with O7+/O6+>1, although four of the complex ejecta contained material with O7+/O6+>1. All of the magnetic clouds caused geomagnetic storms. Three complex ejecta produced no geomagnetic storms. The other two complex ejecta produced geomagnetic storms indirectly: one by driving a shock into the rear of a magnetic cloud and the other by amplifying southward fields in its leading edge and interaction region. Most of the magnetic clouds were associated with a single solar source, but nearly all of the complex ejecta could have had multiple sources. We find evidence in the solar observations that some of the complex ejecta could have been produced by the interaction of two or more coronal mass ejections (CMEs). At least three CMEs might have interacted to produce a large complex ejection that arrived at 1 AU on May 4, 1998. This complex ejection was overtaking and interacting with a magnetic cloud. We discuss several hypotheses concerning the structures and origins of complex ejecta, including the likely possibility that some complex ejecta are formed by a series of interacting CMEs of various sizes. ¿ 2001 American Geophysical Union