We investigate the theoretical uncertainties of P-wave charmonium deca
ys into two pions, chi(cJ) --> pi(+)pi(-), pi(0) pi(0). Constraining t
he pion distribution-amplitude from the recent precise data on F-pi ga
mma(Q(2)), we find an order-of-magnitude discrepancy between data and
prediction. The-disagreement persists even after inclusion of transver
se degrees of freedom and Sudakov suppressions. We propose the colour-
octet mechanism as the solution to the puzzle. The colour-octet decay
contribution arising from the higher Fock component \c(c) over barg$]
of the chi(cJ) wave function is actually not power suppressed with res
pect to the usual colour-singlet decay arising from the dominant \<c(c
)over bar>] Fock state. An explicit calculation yields an agreement wi
th the data for a very reasonable value for the single extra non-pertu
rbative parameter.