The wide variation in the 3He/4He ratio in solar energetic particle events is most plausibly interpreted in terms of two distinct acceleration mechanisms for helium nuclei, one of which favors 3He. 3He-rich events are associated with impulsive energetic electron events which have (at least) two distinct acceleration mechanisms. Logically one of these should accelerate both 3He and electrons. Based on the electron energy spectrum from 3He-rich events, Kahler et al (1987) proposed that the acceleration region should be the high corona, possibly as high as 2 solar radii. Two further observations provide information on where the 3He acceleration might occur. First is the remarkable upper limit on the total 3He fluence (Ho et al, 2005, 2008) which strongly suggests a model where the acceleration process accelerates the majority of 3He ions within a finite reservoir, which can only be a high coronal loop structure. The second is that at quiet times, at 1AU, the 3He/4He ratio is 6-60 times enhanced over the corresponding slow solar wind value (Desai et al, 2006). Both observations indicate that the acceleration which favors 3He, is occurring quasi-continuously in the high coronal structure, with some leakage into the interplanetary medium. Eventually the coronal structure is disrupted and the trapped population is either released into the interplanetary medium or dumped into the "flare" site to provide both energy and seed particles for further acceleration, but by a different process which does not preferentially accelerate 3He. In the former case there will not be a significant flare, but a 3He-rich event, an impulsive electron event, and perhaps a fast jet from one of the footpoints of the trapping structure. In the latter case there will be a significant flare, with a major coronal mass ejection, probably driving a shock, which produces a non-3He-rich event and a much longer and larger energetic particle event.