In the wake of its 2012 report on the biomedical workforce (see Nature 492, 167; 2012), the US National Institutes of Health (NIH) is again seeking proposals for computational models of that workforce, with the aim of tracking and mitigating long-term trends that threaten its size and diversity. Such trends include the rising number of biomedical-PhD holders who seek jobs outside academia and the lower numbers of female scientists who stay in academia after PhDs. Michael Sesma, a programme director for the NIH's National Institute of General Medical Sciences, says that the agency needs models of dynamics, such as how people make career decisions. It will use them to develop grant and other programmes to address problem areas.