PREVIOUS ADVICE 
Pretty much everyone agrees that international scientific cooperation is a good thing. Yet, getting it done can be difficult, especially the funding part. Funding international collaborations is difficult for two reasons: They're collaborative, and they're international. Both aspects present problems.
For reasons of tradition and especially of accountability, scientific grants are typically disbursed to a single institution, usually the principal investigator's (PI's) home institution. If your collaborator is at the same institution you are, collaborations are easy. But, if your collaborator is at a different institution, things are more complicated.
And then there's the international angle: If your collaborator happens to be in a different country, things are more difficult still, because most governmental scientific agencies have a mission to advance a national scientific agenda. The National Science Foundation's (NSF's) core mission is to advance U.S. science and engineering. At the National Institutes of Health (NIH), the mandate is much broader--to improve human health--so NIH is better able to fund research overseas than NSF is. This difference is important but perhaps not as profound as it seems, because NIH also has conditions on funding overseas research, and NSF recognizes that international collaborations are crucial to advancing U.S. science and works to promote them wherever possible. The bottom line is that at both agencies, some things work and others don't. The key to funding international collaborations with money from NIH and NSF is figuring out how the system works and making it work for you.
Of course, NIH and NSF are not the only possible funding sources. Many other government agencies fund scientific research. And then there is the private-nonprofit sector, a legion of (usually disease-specific) philanthropies determined to reduced the incidence of or eradicate particular medical conditions and diseases. But my space is limited, so for present purposes I'll focus on NIH and NSF.
National Institutes of Health
Any discussion of international research funding at NIH must start with the John E. Fogarty International Center for Advanced Study in the Health Sciences (FIC), the only NIH institute or center that exists specifically to fund international research. Yet the pool of money disbursed by FIC is fairly small--about $12 million for new grants in 2004--and the focus on international health disparities means that most of that money goes toward major health problems, such as HIV/AIDS, malaria, emerging infectious diseases, environmental health, and tuberculosis, in the developing world. That's a very good thing, and it's a very important mission, but it's not helpful if your research is in a different area.
But, if it fits, FIC may prove an excellent resource. One FIC program of possible interest is the Global Health Research Initiative Program for New Foreign Investigators, established in 2002 to support the reentry of NIH-trained foreign investigators into their home countries. Although limited in scope, it's an excellent program for those who qualify.
Of broader interest is FIC's Collaboration Award, which uses the R03 mechanism to pay the costs of collaborations between U.S. investigators and their research partners in the developing world. This is a supplemental award open to NIH-funded investigators. Because it is not intended to pay for core research activities, it is limited to $32,000 a year in direct costs.
Just because FIC has "international" in its name doesn't mean that you should limit your search to FIC. The National Institute of General Medical Sciences (NIGMS) alone has a 2004 budget of $1.9 billion, nearly 40 times the budget of FIC, yet NIGMS provides only about 10% of all NIH research grants. NIH is a big place.
Marcus Rhoades, chief of NIGMS's Genetic Mechanisms Branch, emphasized in a recent e-mail exchange that investigators from other countries are free to apply for NIH grants; in principle, you and your collaborator could obtain a matched pair of NIH grants. Or your collaborator could obtain funding from a local agency, with collaboration costs defrayed by NIH or the other agency. An alternative approach is for the domestic researcher--you--to include foreign research expenses--yes, NIH will pay even core research costs for your foreign colleague--as a subcontract. Says Rhoades, "Grants with foreign components have to meet a slightly higher standard, as they are not supposed to duplicate research ongoing in the U.S., and there is some additional paperwork involving the State Department."
Another factor is that reviewers have been known to balk at giving a good priority score to foreign projects, even though they are supposed to focus exclusively on scientific merit. But Rhoades doesn't see this as a serious problem: "I can't recall an instance where a meritorious foreign project was not funded simply because the research was going to be done overseas." Nevertheless, your proposal may not fare as well with a foreign component.
If you do choose to include a foreign component in your proposal, note the following: Facilities and administration for the foreign component of any grant may not exceed 8% of the foreign budget, less equipment. This, in NIH's judgment, is all that's needed to meet the extra costs of compliance with NIH and Department of Health and Human Services requirements.
Rhoades suggests that proposals with foreign components should be written "in the usual manner, although it would be wise to emphasize the unique aspects of the research."
Another tool for the collaborator is what NIGMS calls "collaborative and integrative approaches to research grants" (variants exist across NIH), which is intended to encourage joint ventures. Participating scientists should be funded already; the additional funds are used to facilitate the collaboration. These awards pay up to $300,000.
National Science Foundation
Like NIH, NSF has an entity--the Office of International Science and Engineering (OISE)--that exists to facilitate research overseas. NSF is a great promoter of foreign scientific collaboration, but its narrower mandate generally prevents the agency from funding foreign science directly. But keep reading.
Until recently, OISE provided awards very similar to the collaboration awards provided by the Fogarty International Center: a few thousand dollars per year to defray the extra costs associated with overseas collaborations. OISE no longer makes these kinds of awards; instead, foreign-collaboration expenses should be included as a part of larger research grants submitted to the research directorates, using their deadlines and target dates.
OISE now funds three primary mechanisms, including this one: Planning Visits and Workshops. In addition, OISE supports a wide range of international training programs for scientists at all levels, and it makes postdoctoral fellowship awards specifically for international research.
In general, NSF follows the "sender-side-pays" model: When U.S. scientists go abroad, NSF pays the way. When, say, France sends a collaborator to the United States, France pays. This approach is fine when the partner is France, or some other affluent country, but it works less well when research partners are in developing countries. So--listen up; this is a key point--NSF program officers have some discretion in funding elements of a project that, in the words of Libby Lyons of OISE, "are deemed essential to the successful outcome of the project."
She's making sense; let's keep listening: "Thus, if there is local scientific expertise that is essential to include, a program officer may decide that some consultancy may be appropriate for the foreign scientist. Likewise, if having a foreign student is deemed essential to the project, for example, because s/he speaks the local language or can be in the field all year, a program officer could [provide] some support." Some expenses--tuition at foreign institutions, large amounts of salary for foreign investigators, large-scale foreign training activities--lie well outside NSF's mandate. But the point here is that if they are essential, some foreign expenses can be paid by NSF.
Subcontracts for international work are legal, but they are not common. There's a sort of Catch-22 here: NSF won't pay a lot of money for foreign activities, so most researchers don't bother to apply. In those cases in which larger amounts of foreign support might be justified--usually research activities in the developing world--there are often problems with accountability and financial monitoring.
Expenses associated with foreign collaborations should be included as part of the regular budget. The inclusion of these costs should be justified by means of an "International Justification." As Lyons says, "As in any collaboration, it--the international justification--should describe why the collaboration provides a unique and valuable complement of expertise between the U.S. and foreign scientists." Other benefits of the collaboration--access to unique facilities or resources--should also be emphasized.
As with any NSF proposal, yours should focus on the big picture. Put your project in a larger theoretical context; then get specific. Says Lyons: "Once the general motivation has been set in the larger context, the PI can make the case that an excellent place to study this phenomenon is (e.g.) Ghana. The PI should make clear what the nature of the intellectual collaboration will be and how the strengths of the collaborator complement her or his strengths."
Finally, it's important to remember that the establishment of international ties counts as a "broader impact"--indeed, a very important one. NSF's expanding focus on international activities is reflected in a recent reorganization. As of 1 October 2004, OISE is now located under the agency's head, within the Office of the Director. NSF is limited in the kind of assistance it can supply for foreign collaborations, but the agency will do whatever it can. NSF--and in particular, OISE--exists to help people like you. So give 'em a call.