We sought to understand how GPs make decisions about which patients with symptoms of lung cancer they refer for further investigation. We also examined the extent to which their decisions vary depending on patient or general practice characteristics. We developed an innovative method to study GP behaviour. Over 300 GP participants conducted six ‘consultations’ with simulated patients, delivered through an interactive website. A ‘consultation’ involved first watching a short video of a patient actor reporting a symptom that might indicate lung cancer, and then seeking further information about the patient by typing in questions, or selecting ‘patient note’ options. At the end of the ‘consultation’, GPs recorded their management decisions. The study indicated that when GPs sought pertinent symptoms information, in most cases they sought to investigate lung cancer. However, in 40% of cases they did not seek this information. Moreover even when they did, inequalities in decision making by ethnicity and age remained. The findings for this study have been used to inform an intervention to train future doctors in making decisions when patients present with symptoms that may indicate lung cancer. (see eCREST)