Earlier classification analyses found three types of gamma-ray
bursts (short, long and intermediate in duration) in the BATSE
sample. Recent works have shown that these three groups are also
present in the RHESSI and the Beppo-SAX databases. The duration
distribution analysis of the bursts observed by the Swift
satellite also favors the three component model. In this paper
we extend the analysis of the Swift data with spectral
information. We show, using the spectral hardness and the
duration simultaneously, the maximum likelihood method favors
the three component against the two component model. The
likelihood also shows that a fourth component is not needed.
