There is strong evidence that there are two sub-classes of gamma-ray bursts. The majority of gamma-ray bursts have durations longer than about two seconds and emit most strongly at lower energies. These are the long-soft bursts. A second, smaller sub-class is the short-hard bursts, which have durations less than about two seconds and emit most strongly at higher energies. Swift detects both classes of gamma-ray burst and has been instrumental is measuring their intrinsic properties and determining their progenitors. Swift data has determined the locations, redshift distributions, and afterglow properties of both long-soft and short-hard bursts, allowing a physical understanding of their properties.
Swift has detected more than 45 short-hard bursts and found X-ray afterglows for >70% of them. More than half of these bursts have host galaxy identifications or redshift estimates. The distribution of short-hard bursts within their host galaxies is consistent with these bursta being caused by merging binary neutron stars.
There is growing evidence that long-soft gamma-ray bursts are caused by high-mass stars collapsing in on themselves and becoming supernovae. Swift is studying this by obtaining detailed light curves of Type Ib/c supernova (which are believed to be the progenitors of long-soft gamma-ray bursts) to probe physics of these supernova explosions. However, this picture is still somewhat unclear. In some cases no supernova has been seen down to very faint limits. It is not clear why some bursts appear to have no supernova.