Concerns about the over-prescription of peri-operative fluids, particularly normal saline, culminated in the recent publication of UK national guidelines on fluid prescription during and after surgery. A working group comprising members of the nutrition support team, surgeons, anaesthetists and pharmacists therefore sought to reduce the overall levels of fluid prescription and to limit normal saline usage in our large Teaching Hospital by producing written local fluid prescribing guidelines and holding a series of fluid prescription education sessions for consultants and junior staff. Ideally, the success of such measures would have been determined by studies on fluid balance, body weight and/or measured body water in large numbers of individual patients in a large cluster-randomised controlled trial. However, this would have proved logistically difficult and very costly especially as it is notoriously difficult to rely on the accuracy of daily fluid balance charts in large numbers of patients on busy post-operative surgical wards. We therefore undertook a pragmatic study, comparing historical data on fluid type/volume prescribed (from both individual and ward level pharmacy records), oedema status and clinical outcomes from 2002 with two prospective audits of similar data carried out during 2008 and 2009. Our data showed that in the comparable, elective surgical patients within each audit, there was a decline in total intravenous fluids prescribed over the first 5 post-operative days from 21·1 litres per patient in 2002 to 14·2 litres per patient in 2009 (P<0·05), while pharmacy records showed that the proportion of 0·9% saline supplied declined from 60% to 35% of all fluids supplied to the surgical wards involved, with a concomitant increase in the use of 4%/0·18% dextrose-saline and Hartmann's solution. Alongside these changes in fluid prescribing, the number of patients with clinically apparent oedema declined from 53% in 2002 to 36% in 2009; gut function returned more quickly (6 d in 2002 v. 4 d in 2009, P<0·05) and the length of stay improved from 13 d in 2002 to 10 d in 2009, P<0·05). Although we accept that other factors might have contributed to the observed changes in these clinical parameters, we believe that the measures to reduce fluid and saline administration were the major contributors to these improved clinical outcomes.