Background: Morbidity after major surgery is associated with low oxygen delivery. Haemodynamic therapy aimed at increasing oxygen delivery in an effort to reduce oxygen debt, tissue injury, and morbidity, is controversial. The most appropriate target for this strategy is unclear and might have several off-target effects, including loss of neural (parasympathetic)-mediated cellular protection. We hypothesised that individualised oxygen delivery targeted haemodynamic therapy (goal-directed therapy) in high-risk surgical patients would reduce postoperative morbidity, while secondarily addressing whether goal-directed therapy affected parasympathetic function. Methods: In this multicentre, randomised, double-blind, controlled trial, adult patients undergoing major elective surgery were allocated by computer-generated randomisation to a postoperative protocol (fluid, with and without dobutamine) targeted to achieve their individual preoperative oxygen delivery value (goal-directed therapy) or standardised care (control). Patients and staff were masked to the intervention. The primary outcome was absolute risk reduction (ARR) in morbidity (defined by Clavien-Dindo grade II or more) on postoperative day 2. We also assessed a secondary outcome focused on parasympathetic function, using time-domain heart rate variability measures. Analyses were done on an intention-to-treat basis. The trial was registered with Controlled Clinical Trials (number ISRCTN76894700). Findings: We enrolled 204 patients between May 20, 2010, and Feb 12, 2014. Intention-to-treat analysis of the 187 (92%) patients who completed the trial intervention period showed that early morbidity was similar between goal-directed therapy (44 [46%] of 95 patients) and control groups (49 [53%] of 92 patients) (ARR -7%, 95% CI -22 to 7; p=0·30). Prespecified secondary analysis showed that 123 (66%) of 187 patients achieved preoperative oxygen delivery (irrespective of intervention). These patients sustained less morbidity (ARR 19%, 95% CI 3-34; p=0·016), including less infectious complications. Goal-directed therapy reduced parasympathetic activity postoperatively (relative risk 1·33, 95% CI 1·01-1·74). Interpretation: Achievement of preoperative oxygen delivery values in the postoperative phase was associated with less morbidity, but this was not affected by the use of an oxygen delivery targeted strategy. Reduced parasympathetic activity after goal-directed therapy was associated with the failure of this intervention to reduce postoperative morbidity. Funding: Academy of Medical Sciences and Health Foundation Clinician Scientist Award.