We have compared the value of different variables used in the assessment of blood loss during progressive hypovolaemia and resuscitation under general anaesthesia in anaesthetized pigs. We measured mean arterial pressure (MAP), pulmonary capillary wedge pressure (PCWP), the negative component of the systolic arterial pressure variation (delta Down) and left ventricular end-diastolic area (LVEDa) using echocardiography. Blood was progressively withdrawn (up to 35 ml kg-1 in seven steps) and then reinfused after the same pattern. Regression coefficient (r) and normalized slope (nS) of the regression relationship between each variable and amount of blood loss were determined. The difference between the withdrawal and reinfusion curves was assessed by the area between the curves. We also estimated the minimal loss of blood volume which induced significant changes in each variable compared with that under control conditions during withdrawal of blood (minWBV) and maximal loss in blood volume which induced no significant changes in a variable compared with control conditions during retransfusion (maxRBV). During haemorrhage, MAP decreased (from mean 74 (SD 9) to 31 (5) mm Hg; P < 0.001), delta Down increased (from 1.2 (1.4) to 11.4 (4.2) mm Hg; P < 0.001), PCWP decreased (from 6.2 (2.1) to 0.3 (1.0) mm Hg; P < 0.001) and LVEDa decreased (from 13.8 (2.0) to 5.1 (2.0) cm2; P < 0.01). The highest r values were obtained with MAP and LVEDa, and the highest nS value with delta Down. The least difference between withdrawal and reinfusion was with LVEDa, the lowest values of minWBV were with PCWP and LVEDa, and the highest value of maxRBV was obtained with PCWP. During progressive haemorrhage under general anaesthesia, LVEDa was an accurate variable for assessment of blood volume loss, delta Down contributed no further information compared with MAP, and PCWP was the most reliable variable for assessing return to baseline blood volume.