Shell-type supernova remnants have long been suggested to be sites for cosmic ray acceleration below 100 TeV, mainly on the basis of general energetics arguments. We know from their synchrotron, radio and X-ray emission that electrons are accelerated up to TeV energies. There is, however, no solid evidence for proton acceleration. A possible signature of proton acceleration would be the spectrum of neutral pi decay from collisions of cosmic ray protons and nearby matter like high density molecular clouds. A number of shells have been observed by satellites at 0.1-10 GeV energies and by gamma-ray telescopes at TeV energies; the energy region in between still has to be pinned down experimentally. The origin of this radiation remains uncertain, due to the contamination of gamma-rays produced by leptonic processes (Bremsstrahlung or Inverse-Compton). A low threshold and high sensitivity will allow spectral studies in the range above 10 GeV, where the different mechanisms are expected to show different spectral shapes; it also will allow to pin down the exact positions of possible spectral cutoffs. Plerions or Pulsar-Wind Nebulae are SNRs in which a pulsar wind injects energy into its surroundings. A bubble is inflated out to a radius where it is confined by the expanding shell, as already suggested for the Crab Nebula. Particle acceleration is expected in the wind termination shock. A precise measurement of the spectrum in the energy range below 100 GeV is crucial to constrain the model parameters and to ascertain if another source of photons is necessary, possibly Bremsstrahlung from dense regions of gas.
The observation of gamma-ray pulsars in the GeV domain is of special interest: in this range, the EGRET pulsar spectra cut off due to limited sensitivity; observations of the differential spectra of gamma-ray pulsars in this energy domain gives us clues about the different proposed models (polar cap and outer gap), which explain the gamma-ray emission in neutron stars. The two models predict different cut-off energies: below 40 GeV for the polar cap, up to 100 GeV for the outer gap model. Moreover, many of the unidentified sources recorded in the 3rd EGRET catalogue are believed to be radio-quiet Geminga-like pulsars. Besides the EGRET sources all radio pulsars are expected to lose part of their rotational energy through gamma-ray emission. The large collection area and low energy threshold allow the detection of such possibly faint emission. MAGIC also is sensitive to the weak gamma-ray flux expected from millisecond pulsars, which are predicted to have spectra extending up to a few hundred GeV.