We characterize the tails of the probability distribution functions fo
r the solution of Burgers' equation with Gaussian initial data and its
derivatives partial derivative(k) upsilon(X,t)/partial derivative X(k
), k=0,1,2,.... The tails are ''stretched exponentials'' of the form P
(theta)proportional to exp[-(Re)(-p)t(q) theta(r)], where Re is the Re
ynolds number. The exponents p, q, and r depend on the initial spectru
m as well as on the order of differentiation, k. These exact results a
re compared with those obtained using the mapping closure technique. (
C) 1995 American Institute of Physics.