Can the clinician and the statistician be friends? Andrew Vickers argues that they must work together more closely in order to conduct reliable, useful research that improves patient care.