A new evaluation of the constraint on the number of light neutrino spe
cies (N-nu) from big bang nucleosynthesis suggests a discrepancy betwe
en the predicted light element abundances and those inferred from obse
rvations, unless the inferred primordial He-4 abundance has been under
estimated by 0.014 +/- 0.004 (1 sigma) or less than 10% (95% C.L.) of
He-3 survives stellar processing. With the quoted systematic errors in
the observed abundances and a conservative chemical evolution paramet
rization, the best fit to the combined data is N-nu = 2.1 +/- 0.3 (1 s
igma) and the upper limit is N-nu < 2.6 (95% C.L.). The data are incon
sistent with the standard model (N-nu = 3) at the 98.6% C.L.