I don't think I've ever read a succinct explanation of what makes
an option offense viable, so here it is in a word: mismatches. 
By option offense, I mean a system geared around option plays of
the type where a player (usually a quarterback) with the ball,
usually while "sliding" down the line of scrimmage, makes a
choice to either keep the ball or hand or pitch it to another
back.
    Most offenses pit speed against speed and strength against
strength.  The strongest players, in the line, block their
counterparts in the defensive line and linebackers.  The fastest
players attempt to beat their opponents by using speed running
with the ball from the backfield or going out for forward passes. 
A team that matches up player for player (at least on part of the
field, say) can be effective beating the opposition straight up
that way.
    The option, by contrast, creates mismatches, pitting speed
against strength and strength against speed.  The offensive line
sacrifices blocking against some players of the defensive front,
and are released for duty downfield.  Meanwhile the backs use
their options and their speed against the unblocked linemen and
linebackers to produce 3-on-2 or 2-on-1 situations, or otherwise
just beat the heavies.
    Meanwhile the offensive linemen who are freed at the line of
scrimmage by the blocking scheme knock off players in the
defensive secondary.  In most offenses, the speed of the
defensive backs allows them to beat such downfield blocks. 
However, the option action in the backfield leaves safeties and
corners off balance, freezing them long enough for the linemen to
wipe them out.  When a big lineman can make good contact on a DB
the block is usually very effective, not just a brush -- which
helps when the rules forbid blocking below the waist downfield.
    So option schemes may make give a team that's overmatched at
all positions a chance to make hay on offense, although they're
not necessarily the best choice for a physically superior team,
or a team with a few offensive stars who are hard to cover.

Robert Goodman, April 2000