Previous studies have reported large between-subject variations in the time to exhaustion during intermittent running at the velocity at V(.-)O (2max) (vV(.-)O (2max)). This study aimed to determine which physiological factors contribute to this variability. Thirteen male runners (age 38.9 +/- 8.7 years) each completed five treadmill running tests; two incremental tests to determine V(.-)O (2max), vV(.-)O (2max), the lactate threshold velocity (vLT) and the running velocity--V(.-)O (2) relationship; the third test to determine the time to exhaustion during continuous running at vV(.-)O (2max) (t (lim)cont); the fourth to determine the maximal accumulated oxygen deficit (MAOD); the fifth to determine the time to exhaustion during intermittent running at vV(.-)O (2max) (t (lim)int). Relief intervals during the intermittent test were run at 70 % vV(.-)O (2max). The vLT-vV(.-)O (2max) difference was significantly correlated with t (lim)int (r = - 0.70; p = 0.007). The correlation coefficient increased to r = - 0.83 (p < 0.001) when the difference between the relief interval velocity and the vLT was deducted from the vLT-vV(.-)O (2max) difference (theoretically representing the net depletion of the MAOD during each work/relief interval cycle). The main finding of this study was that 49 % of the variance in t (lim)int was explained by the vLT-vV(.-)O (2max) difference, compared to 74 % for t (lim)cont. However, a further 20 % of unique variance in t (lim)int could be explained with the inclusion of the relief interval velocity-vLT difference. Theoretically, runners with the largest relief interval velocity-vLT difference will replete their anaerobic capacity to a greater extent during each relief interval, thereby increasing time to exhaustion.