There is increasing evidence that intraoperative fluid therapy decisions may influence postoperative outcomes. In the past, patients undergoing major surgery were often administered large volumes of crystalloid, based on a presumption of preoperative dehydration and nebulous intraoperative 'third space' fluid loss. However, positive perioperative fluid balance, with postoperative fluid-based weight gain, is associated with increased major morbidity. The concept of 'third space' fluid loss has been emphatically refuted, and preoperative dehydration has been almost eliminated by reduced fasting times and use of oral fluids up to 2 h before operation. A 'restrictive' intraoperative fluid regimen, avoiding hypovolaemia but limiting infusion to the minimum necessary, initially reduced major complications after complex surgery, but inconsistencies in defining restrictive vs liberal fluid regimens, the type of fluid infused, and in definitions of adverse outcomes have produced conflicting results in clinical trials. The advent of individualized goal-directed fluid therapy, facilitated by minimally invasive, flow-based cardiovascular monitoring, for example, oesophageal Doppler monitoring, has improved outcomes in colorectal surgery in particular, and this monitor has been approved by clinical guidance authorities. In the contrasting clinical context of relatively low-risk patients undergoing ambulatory surgery, high-volume crystalloid infusion (20-30 ml kg(-1)) reduces postoperative nausea and vomiting, dizziness, and pain. This review revises relevant physiology of body water distribution and capillary-tissue flow dynamics, outlines the rationale behind the fluid regimens mentioned above, and summarizes the current clinical evidence base for them, particularly the increasing use of individualized goal-directed fluid therapy facilitated by oesophageal Doppler monitoring.