Are you an aspiring scientist with quantitative chops and a strong desire for a faculty career? Then your best bet would be to take up mathematics. Or, perhaps you would prefer to have a Ph.D.-level career outside of academia. In that case, get your doctorate in computer science or engineering. Do you want to spend several years in low-paying postdoc positions with only a small hope of eventually scoring a tenure-track professorship? If so, the life sciences career track is for you.
Yes, there's some hyperbole in that last sentence. Plenty of people still find good, rewarding academic careers in the life sciences. But, according to a Science Careers analysis of some 2008 data published in the National Science Board's (NSB's) 2012 Science and Engineering Indicators  (S&E Indicators), mathematics looks like an especially promising discipline for those aspiring to an academic career, whereas the life sciences look particularly poor for those with academic ambitions.
There are two parts to this simple analysis. First, by comparing the number of Ph.D. graduates in a field each year to the number of people in faculty jobs in those fields—both using U.S. data—we can estimate what proportion ends up in academic careers. Second, by comparing the number of faculty in the field to the number doing a postdoc—which besides being a training phase also serves as a holding pattern for scientists waiting and hoping for faculty jobs—we can get an idea of whether Ph.D. scientists have access to other career options they deem acceptable.
Overall—in all scientific and engineering fields—the ratio of Ph.D.s awarded annually (41,180 in 2008) to the number of science and engineering faculty (199,500 in the same year) gives a number (in this case, 0.21). If we make an arbitrary assumption about the average length of a scientific career—let's assume it lasts 20 years—then we can conclude that about 24% of all science Ph.D. recipients work in academia. It's a rough estimate that depends on a number of large assumptions, but it is useful at least for comparisons among fields.
How do the odds look field by field? The answers are in the table below. By a considerable margin, mathematics Ph.D. recipients are the most likely to hold a faculty post, with a Ph.D.-to-faculty ratio of just 0.12—a little more than eight current faculty members for each new Ph.D. awarded. Assuming a 20-year career, this ratio implies that about 43% of Ph.D. recipients in mathematics can expect to hold faculty positions. The next best field for academic careers is the physical sciences, where the ratio is 0.15, implying that about 33% of Ph.D. recipients can expect to serve on college and university faculties.
In the three other categories—biological science (including biomedical), computer science, and engineering—Ph.D. recipients are much less likely to hold faculty positions than they are in math and the physical sciences. In the biological sciences, the Ph.D.-to-faculty ratio is 0.24, indicating, with the same career-length assumptions, that about 21% of those with Ph.D.s in these fields can expect faculty jobs. In computer science, the ratio is about 0.28, suggesting about an 18% faculty yield. In engineering, the ratio is 0.36, which means that just 14% of engineering Ph.D. recipients end up in faculty jobs.
Of course, these latter fields—computer science and engineering—aren't primarily academic fields, so it's not surprising that so many Ph.D. computer scientists and engineers are employed outside academia. They have other opportunities—that's a good thing.
So, we need another metric to measure the health of a field, and the S&E Indicators offer a possibility: the ratio of the number of current faculty to the number of postdocs. A field with a higher share of postdocs is likely to have stiffer competition for scarce academic careers, and a large proportion of postdocs very likely indicates that the field offers few compelling alternatives to the academic track. The table below ranks fields in terms of the number of postdocs per faculty position, from best (lowest) to worst (highest):
Again, mathematics tops the list, with the fewest postdocs per faculty member. Computer science does almost as well. Engineering, physical sciences, and life/med sciences are clustered together at the other end of the scale, but the life/med category is significantly worse than the others.
Combining these data makes mathematics stand out. It's a unique field in that a large proportion of graduates end up in faculty posts—the jobs graduate school prepares them for—and yet its low postdoc numbers suggest that most graduates are placed in real jobs soon after receiving their Ph.D.s; few mathematicians, it seems, get stuck in the postdoctoral holding pattern.
Computer science also seems healthy, with a relative few finding work in academia but a very good faculty-to-postdoc ratio, suggesting that Ph.D. computer science graduates have very strong nonacademic career prospects. In terms of the postdoc metric, engineering and physical sciences don't appear particularly strong—but the life/med discipline is the weakest. Here, only a small percentage are in faculty jobs and a large proportion are in the postdoc holding pattern, where they continue to train for faculty jobs that, statistics show, few will ever get.
* Compared to some other sources, the number of postdocs reported by NSB seems quite low, especially in biomed.