The aims of this study were 1) to evaluate whether subjects suffering from acute mountain sickness (AMS) during exposure to high altitude have signs of autonomic dysfunction and 2) to verify whether autonomic variables at low altitude may identify subjects who are prone to develop AMS. Forty-one mountaineers were studied at 4,559-m altitude. AMS was diagnosed using the Lake Louise score, and autonomic cardiovascular function was explored using spectral analysis of R-R interval and blood pressure (BP) variability on 10-min resting recordings. Seventeen subjects (41%) had AMS. Subjects with AMS were older than those without AMS (P < 0.01). At high altitude, the low-frequency (LF) component of systolic BP variability (LF(SBP)) was higher (P = 0.02) and the LF component of R-R variability in normalized units (LF(RR)NU) was lower (P = 0.001) in subjects with AMS. After 3 mo, 21 subjects (43% with AMS) repeated the evaluation at low altitude at rest and in response to a hypoxic gas mixture. LF(RR)NU was similar in the two groups at baseline and during hypoxia at low altitude but increased only in subjects without AMS at high altitude (P < 0.001) and did not change between low and high altitude in subjects with AMS. Conversely, LF(SBP) increased significantly during short-term hypoxia only in subjects with AMS, who also had higher resting BP (P < 0.05) than those without AMS. Autonomic cardiovascular dysfunction accompanies AMS. Marked LF(SBP) response to short-term hypoxia identifies AMS-prone subjects, supporting the potential role of an exaggerated individual chemoreflex vasoconstrictive response to hypoxia in the genesis of AMS.