Introduction Low energy availability (EA) may impede adaptation to exercise, suppressing reproductive function and bone turnover. Exercise energy expenditure (EEE) measurements lack definition and consistency. This study aimed to compare EA measured from moderate and vigorous physical activity from accelerometry (EEEmpva) with EA from total physical activity (EEEtpa) from doubly-labelled water in women. The secondary aim was to determine the relationship of EA with physical fitness, body composition by DXA, heartrate variability (HRV) and eating behavior (brief eating disorder in athletes-questionnaire, BEDA-Q). Methods Prospective, repeated measures study, assessing EA measures and training adaptation during 11-month basic military training. 47 women (23.9 ±2.6 years) completed 3 consecutive 10-d assessments of EEEmvpa, EEEtpa and energy intake (EI). EA measures were compared using linear regression and Bland-Altman analyses; relationships of EA with fat mass, heartrate variability, 1.5-mile run times and BEDA-Q were evaluated using partial correlations. Results EA from EEEmvpa demonstrated strong agreement with EA from EEEtpa across the measurement range (R2=0.76, r=0.87, p<0.001) and was higher by 10 kcal/kg FFM/d. However, EA was low in absolute terms due to underreported EI. Higher EA was associated with improved 1.5 mile run time (r=0.28, p<0.001) fat mass loss (r=0.38, p<0.001) and lower BEDA-Q score (r=–0.37, p<0.001) but not HRV (all p>0.10). Conclusion Accelerometry-based EEE demonstrated validity against DLW during multi-stressor training, the difference representing 10 kcal/kg FFM/d EEE from non-exercise activity. Beneficial physical but not autonomic adaptations were associated with higher EA. EAmvpa and BEDA-Q warrant consideration for low EA assessment and screening.