Increasingly restrictive red blood cell transfusion strategies have led to growing numbers of patients being discharged from hospital with moderate anaemia, and doubts about the effects of this trend on the outcomes of patients with anaemia are growing. One advantage of doubts is that they lead to variation in practice, which, in theory, allows the study of routinely collected data on the effects of an intervention (transfusion) on patient-relevant outcomes. To do this in a reliable way, we need solid data on outcomes and on all the potentially relevant clinical characteristics that are associated with both the decision to transfuse and the outcomes. Currently, such data are not available. As part of routine practice, clinicians should be encouraged to document the arguments for their treatment decisions and the intended outcomes. This will allow learning from daily practice in the field of transfusion medicine and in other health care disciplines.