The study tested the hypothesis that running velocity corresponding to the anaerobic threshold (VAT) would more accurately approximate the actually measured marathon race velocity (VM) than would running velocity corresponding to the so-called onset of blood lactate (4 mM) accumulation (VOBLA). The VAT (4.57 m X s-1) well approximated the VM (4.49 m X s-1), whereas the VOBLA (5.30 m X s-1) differed significantly from the VM. In addition, the VAT (r = 0.781) correlated with VM to a greater extent than did the VOBLA (r = 0.682). When the VAT (X1) was combined with delta % maximum O2 consumption (VO2max) (%VO2max at the OBLA minus %VO2max at the AT; X2) and VO2max (ml X min-1 x kg-1; X3), variation in the VM accounted for increased profoundly from 61 to 88%. Thus one of the useful equations formulated with high predictive accuracy was VM (m X s-1) = 1.312X1 + 0.0346X2 - 0.00993X3 - 1.272. Our study demonstrates that the anaerobic threshold (AT) is more closely associated with marathon running performance and that the degree of the association is raised when delta %VO2max and/or VO2max are combined as additional information.