The theoretical predictions of big bang nucleosynthesis (BBN) are dominated
by uncertainties in the input nuclear reaction cross sections. In this pap
er we examine the impact on BBN of the recent compilation of nuclear data a
nd thermonuclear reactions rates by the NACRE collaboration. We confirm tha
t the adopted rates do not make large overall changes in central values of
predictions, but do affect the magnitude of the uncertainties in these pred
ictions. We therefore examine in detail the uncertainties in the individual
reaction rates considered by NACRE. When the error estimates by NACRE are
treated as 1 sigma limits, the resulting BBN error budget is similar to tho
se of previous tabulations. We propose two new procedures for deriving reac
tion rate uncertainties from the nuclear data: one which sets lower limits
to the error, and one which we believe is a reasonable description of the p
resent error budget. We propagate these uncertainty estimates through the B
BN code, and find that when the nuclear data errors are described most accu
rately, the resulting light element uncertainties are notably smaller than
in some previous tabulations, but larger than others. Using these results,
we derive limits on the cosmic baryon-to-photon ratio eta, and compare this
to independent limits on eta from recent balloon-borne measurements of the
cosmic microwave background radiation (CMB). We discuss means to improve t
he BBN results via key nuclear reaction measurements and light element obse
rvations. (C) 2001 Published by Elsevier Science B.V.