It is a near-unavoidable paradox. Scientists pursue a career in science because they love research, but the more successful they become, the farther they find themselves from the bench. Instead of collating data, they get bogged down in the minutiae of management. What's worse, few find themselves prepared for such a role, even though running a lab, getting grants and cementing alliances all depend on management skills.

So where does one get such skills? Some people are naturals; most muddle through. But a few scientists who cut their scientific and management teeth in industry, and who have now returned to run academic institutions, are suggesting that some formal training may be in order.

Richard Sykes, former chairman of GlaxoSmithKline and now rector of Imperial College London, feels that there is a need for specialized programmes in the management of science, technology and engineering. Compared with an MBA, these courses would teach skills targeted at science and would create a smoother career path from student to postdoc to manager (see http://www.nature.com/naturejobs/channels/graduate/int_sir-richard-sykes.html). And Gail Naughton, dean of the College of Business Administration at San Diego State University, has launched a combined MBA–PhD programme that has internship, thesis and business-plan components (see Nature 430, 706-707; 2004).

Clearly, there is no one right way to teach scientific management, just as there is no one archetypal scientific manager. The different demands of small labs, large multidisciplinary efforts and multi-site initiatives all require different approaches. But putting management courses into the training mix of a PhD, then honing these skills during a postdoc, would help young scientists when they become principal investigators — allowing them to be ready for a role that their scientific skills will make inevitable.