On ten top-level Kenyan marathon runners (KA) plus nine European controls (EC, equivalent to KA), we measured maximal oxygen consumption (VO2max) and the energy cost of running (Cr) on track during training camps at moderate altitude, to better understand the KA dominance in the marathon. At each incremental running speed, steady-state oxygen consumption (VO2) was measured by telemetric metabolic cart, and lactate by electro-enzymatic method. The speed requiring VO2 = VO2max provided the maximal aerobic velocity (νmax). The energy cost of running was calculated by dividing net VO2 by the corresponding speed. The speed at lactate threshold (ν(ΘAN)) was computed from individual Lâ(b) versus speed curves. The sustainable VO2max fraction (Fd) at ν(ΘAN) (F(ΘAN)) was computed dividing nu(ΘAN) by νmax. The Fd for the marathon (Fmar) was determined as Fmar = 0.92 F(ΘAN). Overall, VO2max (64.9 ± 5.8 vs. 63.9 ± 3.7 ml kg(-1) min(-1)), νmax (5.55 ± 0.30 vs. 5.41 ± 0.29 m s(-1)) and Cr (3.64 ± 0.28 vs. 3.63 ± 0.31 J kg(-1) m(-1)) resulted the same in KA as in EC. In both groups, Cr increased linearly with the square of speed. F(ΘAN) was 0.896 ± 0.054 in KA and 0.909 ± 0.068 in EC; Fmar was 0.825 ± 0.050 in KA and 0.836 ± 0.062 in EC (NS). Accounting for altitude, running speed predictions from present data are close to actual running performances, if F(ΘAN) instead of Fmar is taken as index of Fd. In conclusion, both KA and EC did not have a very high VO2max, but had extremely high Fd, and low Cr, equal between them. The dominance of KA over EC cannot be explained on energetic grounds.