In patients with haemophilia, the development of neutralizing allo-antibodies ('inhibitors') while receiving the deficient clotting factor is relatively common during the patients' initial treatment. Among treated patients with haemophilia who do not develop inhibitors early on, the later incidence is considerably lower. Therefore, the evaluation of potential risk factors for their tendency to give rise to inhibitors is best performed separately in previously untreated and previously treated patients. We discuss potential implications of study design and analysis choices on the validity of inferences from studies assessing inhibitor incidences.