You're a new investigator and you are trying to decide how best to get money to support your new lab. You may be weighing the relative risks and rewards of submitting as your first NIH research project grant application an R21 versus an R01. This is actually one of the easist decisions you will ever make as a new PI.
If we weigh the costs and benefits of an R01 versus an R21 grant application, you will see that only in very particular circumstances does it make sense for a new PI to spend any of her time writing R21s.
R01s have the following benefits over R21s: (1) Limited to five years instead of two years. (2) Modular budget limited to $250,000 per year for all five years, instead of only $137,500 per year for two years. (3) Competitive renewal permitted.
R21s have the following ostensible benefits over R01s: (1) 15 page application, instead of 25. (2) Although R21 Funding Opportunity Announcements differ in their exact focus, one commonality is that much less preliminary data than for an R01 is explicitly demanded. Are these benefits real?
Well, it takes nearly as much time to prepare an excellent R21 application as an R01, so the shorter page limit is of minimal relevance. And "less (actually 'no') preliminary data required", while theoretically fantastic for a new PI, is actually pretty close to bullshit. Unless the proposed studies are patently feasible (and, thus, probably boring and getting hit on significance), study sections almost certainly will ding the absence of preliminary data as a "feasibility" issue, but just without using the magic verboten words "preliminary data". When all is said and done, however, the sheer amount of preliminary data required to satisfy an R21 study section is somewhat less than that typically required for an R01.
Now, figure out which of the following two populations you are a member of:
Population #1: You have sufficient space and start-up resources to run your lab with two post-docs (or equivalent) and a technician (or post-graduate researcher) for at least three years without any grant income.
Population #2: You do not have sufficient space and start-up resources to run your lab with two post-docs (or equivalent) and a technician (or post-graduate researcher) for at least three years without any grant income.
If you are in population #1, you should focus your time and efforts solely on getting a first R01 funded, and forget about R21s. You should have enough start-up resources--if you plan and manage effectively--to generate preliminary data for at least two credible scientifically distinct R01 applications. An R21 application is too little money for too little duration in return for all that effort, and does not provide a competitively renewable long-term expected future flow of grant support that is essential for sustained viability of your independent career. This is essential for tenure, and if you look at success rates for competitive renewals you will see why a competitive renewal application has a greater expected future value than a new application. Since you have sufficient start-up resources to keep hammering experimentally on two distinct R01s for at least three years, you can afford the risk of both R01s requiring a second resubmission (i.e., reaching "A2" status) before one of them hits.
If you are in population #2, the decision becomes a little more tricky, and depends on how much start-up resources you can command. If you have essentially none, you are pretty much stuck with R21 applications, because you have no way to generate preliminary data sufficient for an R01. Once you manage to get an R21, that gives you the resources to leverage into an R01.
If you have only enough for two or three person years of experimental effort (about half of that required to place you in population #1), then it requires a quantitative weighing of risk and benefit. The danger of going for R01s only, is that you may end up running out of start-up resources before getting two distinct R01 applications to A2s, and getting one awarded. You might be fortunate enough to get one sooner, and thus be ok, but you would be taking a bigger risk than someone in population #1.
If you go for one R01 and one R21, on the other hand, you diversify your risk better, and the R21 gives you a shot of getting some substantial funding even in the absence of a pile of preliminary data sufficient for an R01. And then you can leverage the resources provided by an R21--essentially an extra infusion of start-up--into an R01 (or a second one if you are fortunate).