In the intravenous iron therapy to treat iron deficiency anaemia in patients undergoing major abdominal surgery (PREVENTT) trial, the use of intravenous iron did not reduce the need for blood transfusion or reduce patient complications or length of hospital stay. As part of the trial protocol, serum was collected at randomisation and on the day of surgery. These samples were analysed in a central laboratory for markers of iron deficiency. We performed a secondary analysis to explore the potential interactions between pre-operative markers of iron deficiency and intervention status on the trial outcome measures. Absolute iron deficiency was defined as ferritin <30 μg.l-1 ; functional iron deficiency as ferritin 30-100 μg.l-1 or transferrin saturation < 20%; and the remainder as non-iron deficient. Interactions were estimated using generalised linear models that included different subgroup indicators of baseline iron status. Co-primary endpoints were blood transfusion or death and number of blood transfusions, from randomisation to 30 days postoperatively. Secondary endpoints included peri-operative change in haemoglobin, postoperative complications and length of hospital stay. Most patients had iron deficiency (369/452 [82%]) at randomisation; one-third had absolute iron deficiency (144/452 [32%]) and half had functional iron deficiency (225/452 [50%]). The change in pre-operative haemoglobin with intravenous iron compared with placebo was greatest in patients with absolute iron deficiency, mean difference 8.9 g.l-1 , 95%CI 5.3-12.5; moderate in functional iron deficiency, mean difference 2.8 g.l-1 , 95%CI -0.1 to 5.7; and with little change seen in those patients who were non-iron deficient. Subgroup analyses did not suggest that intravenous iron compared with placebo reduced the likelihood of death or blood transfusion at 30 days differentially across subgroups according to baseline ferritin (p = 0.33 for interaction), transferrin saturation (p = 0.13) or in combination (p = 0.45), or for the number of blood transfusions (p = 0.06, 0.29, and 0.39, respectively). There was no beneficial effect of the use of intravenous iron compared with placebo, regardless of the metrics to diagnose iron deficiency, on postoperative complications or length of hospital stay.