During the last few years there has been a shift in the paradigm
of solar particle acceleration based on improved measurements of
particles and their correlation with solar phenomena. We now
recognize that the particles in most of the largest events are
accelerated over a large spatial region by the shock wave ahead
of a CME, not in a solar flare. The particles that
are accelerated in association with impulsive flares have
unusual 
He-rich, Fe-rich abundances resulting from
electron-beam induced resonant wave-particle interactions in the flare.
In fact, particle abundances and ionization states have become a
tool that can distinguish the two mechanisms of acceleration,
even in some events where both mechanisms are present [
Reames 1990b].
The identification of the particle sources has changed our ideas of particle transport. With extended shock acceleration, we no longer need ad hoc concepts like ``coronal diffusion'' to explain the longitude distribution or time profiles, and the particles stream outward from any source with only moderate scattering. In large events the time profiles tell us mostly about the evolution of the source and our connection to it, and almost nothing about particle transport from it.
Having the correct acceleration paradigm is of great practical
importance if one is to predict particle events at Earth. The
highest intensities and longest durations of 1-10 MeV protons are
likely to come from CMEs launched near central meridian. The
highest energy protons, accelerated fairly near the Sun in all
cases, will come from events near 
50
west.
Highly ionizing Fe ions will be especially abundant at 1 AU when large
impulsive flares occur at 
40
-70
west. As a large,
magnetically-complex active region traverses the solar disk, the
intensity, abundances, spectra and time profile of the particles
that might be expected from events in the region will change with time.
We have only begun to understand the physics of particle acceleration in solar events and many questions remain. In the impulsive events the mechanism of acceleration of the intense electron beams is not clear, both electric-field and stochastic mechanisms seem possible. We cannot yet predict the abundances of all the elements in these events and the role of wave cascading is not clear. In the gradual events we do not know the structure, strength and time evolution of the shock far around on the flanks and especially in the corona. We are especially uncertain about the connectivity and topology of the magnetic field lines close behind the shock within the ``ejecta'' and still farther behind where field lines may be ``drawn out'' by the CME [see Reames 1994]. We cannot even predict the abundance variations that occur in different gradual events. Recognizing the approximate site of the acceleration is only the beginning.
It is now clear that the errors of the previous paradigm came
about partly because of an excessive focus on protons. This made
proton-poor events seem small and inconsequential so they were
overlooked while great effort was lavished on fitting proton time
profiles and anisotropies in large events. Much of our new
understanding has come from the study of the abundances of
elements and isotopes and of their ionization states. Not only
do these ions tell us about the conditions in the source plasma
but they have opened a new window on the complex plasma physics
of particle acceleration. The existence of resonant wave
interactions that enhance 
He by many orders of magnitude can
not be inferred from photon observations of flares. Similar
processes may be important in distant sources of astrophysical
interest where direct particle observations are not possible.