PURPOSE: To determine whether measuring heart rate variability (HRV) in a group of septic patients without multiple organ dysfunction syndrome (MODS) made it possible to predict which of them would later develop this syndrome. MATERIAL AND METHODS: We studied 46 septic patients without MODS at the time of admission to an intensive care unit (ICU). During the first 24 hours of admission, a 10-minute electrocardiogram (ECG) was performed and 8 HRV indexes were calculated off-line. Eleven patients later developed MODS (MODS group) during their ICU stay, and 28 did not (non-MODS group). Seven patients were excluded. RESULTS: Although Acute Physiological and Chronic Health Evaluation (APACHE II) scores were similar for both groups, most HRV indices on admission were reduced significantly in the MODS group. Compared with a subset from the non-MODS group (control group, n = 11) paired by age, the MODS group had significantly lower low-frequency spectral components (LF, P =.0128) and mean squared successive differences of R-R intervals (rMSSD) (P =.0473) values. Multivariable logistic regression identified LF as the best predictor of MODS and received operating characteristic (ROC) curves established its cut-off point at 18 ms(2). Mortality rates were 63.6% for the MODS group and 0% for the non-MODS group (P <.0001). CONCLUSIONS: Reduction of HRV on ICU admission may be useful in identifying septic patients at risk for development of MODS.