When the Lockheed YF-22 and Northrop YF-23 were unveiled in 1990, it was
generally believed that the two companies had made different trade-offs
among the various design requirements.  The YF-23 appeared to be optimised
for stealth, with its trapezoidal wings, butterfly tail, and generally
futuristic appearance (the distinct resemblance to the fictional "Firefox"
attracted a lot of comments).  The YF-22, on the other hand, had a more
conventional appearance; although it was obviously designed with stealth in
mind, there was a definite resemblance to the F-15 it was intended to
replace, and the impression was of an aircraft designed for manoeuvrability
first and stealth second.  The YF-22 had thrust-vectoring jet nozzles,
while those of the YF-23 were designed to hide the engines' infrared
signature from below.

In April 1991, the YF-22 was selected for production.  According to the
USAF, neither aircraft showed any clear advantage in either manoeuvrability
or stealth.  The reasons given for the choice were that the Lockheed
aircraft was better designed for maintainability, had more potential for
future development, and was slightly cheaper.

An unconfirmed report has it that one factor was the fact that the YF-23
had its internal AAMs "stacked" in its bays, while the YF-22's missiles
each had a bay to themselves; this meant that, on the YF-23, a malfunction
in one launcher might prevent the launch of another missile in the same
bay.

There remains a popular opinion that the reasons given were bogus, and that
a preference for manoeuvrability over stealth was the real reason for the
choice.  However, there is no obvious reason why the USAF should want to
lie about its reasons, and it seems likely that the external appearance of
the two aircraft wasn't as good a guide to their capabilities as many
people thought.

[From Mike Spick & Barry Wheeler, _Modern American Fighters and Attack
Aircraft_, and magazine reports]