In this work, bioelectrical impedance vector analysis (BIVA) method is used in a sample of haemodialysis patients in stable (without oedema) and critical (hyperhydrated and malnutrition) states, in order to establish the relation between hyperhydration (oedema) and mortality. The measurements obtained were single frequency (50 kHz), tetrapolar (hand-foot) complex impedance measurements (vector components are: resistance R and reactance Xc). The impedance components were standardized by the height H of the subjects, (R/H and Xc/H) to obtain de impedance vector Z/H, that is represented in the RXc plot (abscise R/H, ordinate Xc/H). Measurements were performed on a sample of 74 patients (30 men and 44 women, 18-70 year, body mass index (BMI), 19-30 kg m(-2)) at the Saturnino Lora University Hospital in Santiago de Cuba. The 46 stable patients comprised 28 men and 18 women; the 28 critical patients 16 men and 12 women. The reference population consisted of 1196 healthy adult subjects living in Santiago de Cuba (689 men and 507 women, 18-70 year, BMI 19-30 kg m(-2)). We used the RXc plot with the BIVA method to characterize the reference population using the 50%, 75% and 95% tolerance ellipses. Student's t-test and Hotelling's T2-test were used to analyse the separation of groups obtained by means of clinical diagnosis and those obtained by BIVA. We obtained a significant difference (P < 0.05) in R/H, Xc/H and phase angle (PA) in men as in women between the location of Z/H vectors in the RXc graph and the separation made by the doctors between stable and critical patients. Critical (hyperhydrated) patients were located below the inferior pole of the 75% tolerance ellipse, whereas stable patients were within the tolerance ellipses. Some cases classified as stable by the clinic were classified as hyperhydrated by BIVA with 100% sensitivity and 48% specificity. In conclusion, the BIVA method could be used to classify patients by hydration state and to predict survival. Advantages of the method are its simplicity, objectivity and that it does not require the definition of patient dry weight.