It's 1904. Einstein gets three grants. Annus mirabilis becomes annus satisfactory. It seems obvious, from our present perspective, that for Einstein to write and administer a collection of grants would have destroyed his creativity. The US National Institutes of Health (NIH), however, supports many researchers who are principal investigators of four, five, six or more research grants. Have these grants been awarded without concern for the applicant's scientific creativity? What peer-review committees (called study sections) make these awards? Although giving many grants to one person probably doesn't often result in bad science (although a lack of supervision leading to fraud may be an issue), the real penalty is the loss of ideas. We are giving funds to provide predictable new life at the expense of the unexpected1.

Given a fixed NIH budget, one person can only get their research funded at the expense of someone else. For every grant awarded to someone who already has other NIH grants, one or more people are dismissed. Instead of multiple grants requiring less money because they build on the infrastructure of pre-existing grants, the average value increases with the number of grants held by one person2,3. Needless to say, the losers are most likely to be beginners who do not have an existing laboratory to generate preliminary data. A competitive NIH grant submission has about a 90 per cent chance of getting rejected4, and this high probability encourages smart young people to leave research.

In 1986, more than 96 per cent of grant awards in the category reserved for first-time applicants went to people under 36 years of age. By 1993, this proportion had fallen to less than 29 per cent (ref. 5). In 1994, 38 per cent of all first-time awards went to people between 36 and 40 years old6. This trend is bad for science. Biology has made its greatest advances from the ideas of young independent investigators. The shortage of funds has its greatest effect on those who are not ‘in’, as shown by that ground-breaking study7, done in Sweden, showing that women needed vastly superior credentials, equivalent to two or three extra publications in Nature or Science, to have the same chance as men of getting a postdoctoral fellowship (see page 204). The study showed that men who were not associated with members of the peer-review committees were also treated unfairly.

There is no reason to believe that the United States is different. In 1995, 45 per cent of PhDs in biomedical sciences (US citizens in US universities) were awarded to women, but less than 8 per cent of all research-centre grants were awarded to women6,8. How many women seek funding? Not as many as men, if the Federation of American Societies for Experimental Biology can be taken as representative of the funded biomedical research community9: women constitute only about 13 per cent of its membership.

The NIH appears to be making an effort to maintain balanced reviews: its study sections contain on average 20 per cent women. Coincidentally, about 20 per cent of research projects are awarded to women. The presence of a male majority on study sections, however, could be a significant factor in rating proposals from women. This is particularly important now because the number of female graduates is up from 25 per cent in 1975 to 45 per cent in 1995.

How many grants should a person control? A random search of the CRISP database (http://www.nih.gov/grants/award/crisp.htmtarget=new>http://www.nih.gov/grants/award/crisp.htm) shows at least one principal investigator on 11 different grants, some of which are large programme project grants. Because CRISP is not relational, it is not possible to identify the real grant winners. CRISP also has no information on whether a principal investigator is a co-principal investigator on any other NIH grants, a recipient of grants from other sources, or has any administrative or clinical duties — or even for how much the grants are funded.

The principal investigator with 11 grants, however, has less than half a day to work on each one during a five-day week. This assumes no travel, and no institutional obligations. Should this level of commitment be encouraged while other applicants are consigned to the trash? If we give someone US$2 million a year to perform a clinical study, should we give them six more research grants (NIH lists some people as principal investigators on seven such grants2)? Should we add to that grant supply some training grants that pay for students to work on these projects? In large laboratories most training is done by other students and fellows, not laboratory directors. In compiling data on multiple grants, I received no responses to questions about justification for funding from some of these successful principal investigators. Perhaps they were too busy.

The publicly funded research enterprise is short of money to support independent investigators. Where should it come from? As things stand, it can come either from an increased NIH budget or from a redistribution. (A large increase in funding from Congress seems unlikely, and in any event, given present funding policy, an increased budget would lead to more money going to currently funded investigators.) If, instead, the NIH were to limit the number of grants and to decrease the overhead rates for multiple grants from the same principal investigator, there could be significant changes without a decrease in the number of funded labs. Limiting the number of grants to two for each principal investigator would provide funds for 3,000 new principal investigators. One thing is clear: the NIH must act now to stop the loss of promising young investigators.