We present a model of a freely precessing neutron star, which is then compa
red against pulsar observations. The aim is to draw conclusions regarding t
he structure of the star, and to test theoretical ideas of crust-core coupl
ing and superfluidity. We argue that, on theoretical grounds, it is likely
that the core neutron superfluid does not participate in the free precessio
n of the crust. We apply our model to the handful of proposed observations
of free precession that have appeared in the literature. Assuming crust-onl
y precession, we find that all but one of the observations are consistent w
ith there being no pinned crustal superfluid at all; the maximum amount of
pinned superfluid consistent with the observations is about 10(-10) of the
total stellar moment of inertia. However, the observations do not rule out
the possibility that the crust and neutron superfluid core precess as a sin
gle unit. In this case the maximum amount of pinned superfluid consistent w
ith the observations is about 10(-8) of the total stellar moment of inertia
. Both of these values are many orders of magnitude less than the 10(-2) va
lue predicted by many theories of pulsar glitches. We conclude that superfl
uid pinning, at least as it affects free precession, needs to be reconsider
ed.