Background: Body composition is commonly altered in response to critical illness and can be estimated at the bedside with bioelectrical impedance spectroscopy (BIS). Different electrode configurations may be used to mitigate assumptions of the technique, but the reliability of tetra-polar and octo-polar arrangements has yet to be established. This study aimed to compare both configurations, in a prospective observational study of 17 critically ill survivors and 12 healthy controls. Methods: Weight, supine body length, and BIS on both tetra-polar and octo-polar configured devices were recorded, then repeated 2 days later. Bioelectrical impedance vector analysis was subsequently performed using data from the tetra-polar device at a frequency of 50 kHz. Results: Test-retest agreement was acceptable for the tetra-polar device (intraclass correlation coefficient range, patients: 0.876-0.988 vs controls: 0.983-0.998, P = 0.001). However, lower and wider ranging test-retest intraclass correlation coefficients were obtained with the octo-polar instrument in both groups. Furthermore, there was a difference in the mass/volume of body compartments measured on each device in both patients (P = .017) and controls (P = .045). A change in the composition profile of critically ill males was evident between measurement occasions, which was reflected by a reduction in body weight of 1.6 (1.5) kg (P = 0.001) across the sample over the same period. Conclusions: BIS devices should not be used interchangeably in the clinical setting. The reliability of the tetra-polar instrument was good, but daily fluctuations in body weight may have affected the results.