Introduction We repeated our study of intensified training on a ketogenic low-carbohydrate (CHO), highfat diet (LCHF) in world-class endurance athletes, with further investigation of a "carryover" effect on performance after restoring CHO availability in comparison to high or periodised CHO diets. Methods After Baseline testing (10,000 m IAAF-sanctioned race, aerobic capacity and submaximal walking economy) elite male and female race walkers undertook 25 d supervised training and repeat testing (Adapt) on energy-matched diets: High CHO availability (8.6 g�kg-1 �d-1 CHO, 2.1 g�kg-1 �d-1 protein; 1.2 g�kg-1 �d-1 fat) including CHO before/during/after workouts (HCHO, n = 8): similar macronutrient intake periodised within/between days to manipulate low and high CHO availability at various workouts (PCHO, n = 8); and LCHF (<50 g�d-1 CHO; 78% energy as fat; 2.1 g�kg-1 �d-1 protein; n = 10). After Adapt, all athletes resumed HCHO for 2.5 wk before a cohort (n = 19) completed a 20 km race. Results All groups increased VO 2 peak (ml�kg-1 �min-1) at Adapt (p = 0.02, 95%CI: [0.35-2.74]). LCHF markedly increased whole-body fat oxidation (from 0.6 g�min-1 to 1.3 g�min-1), but also the oxygen cost of walking at race-relevant velocities. Differences in 10,000 m performance were clear and meaningful: HCHO improved by 4.8% or 134 s (95% CI: [207 to 62 s]; p < 0.001), with a trend for a faster time (2.2%, 61 s [-18 to +144 s]; p = 0.09) in PCHO.