One measurable indicator of the state of our physique is our body fat percentage. The popularity of this number may be due to the simple, but incorrect picture it evokes: your body is made of healthy stuff plus fat. We've been conditioned to assume that fat is bad, hence our goal should be to minimize body fat percentage.
Of course, reality is much more complicated than this simple picture. First, while excessive fat is generally correlated with poor health outcomes, your body needs some fat (essential fat) to function properly. Zero body fat percentage implies death. A more appropriate goal might be to maintain a healthy body fat percentage. Unfortunately, nobody really knows what an optimal number may be for any given individual. Evidence suggests wide variation among individuals. Second, the single number "body fat percentage" fails to discriminate between the various types of fat tissue present in the body. It also fails to provide any indication of the distribution of fat within the body. Finally, like any measurement, body fat percentage is subject to errors. Depending on the method of measurement and/or the skill of the technician, estimates can vary wildly.
For these reasons, body fat percentage is probably not a number that anyone should spend a lot of time obsessing over. Nevertheless, it can be an interesting indicator. It's best use may be as a long-term tracker of body composition. By measuring periodically, in the same way each time, one can get some indication of how body composition evolves in response to diet and training strategies, and with age. However, given the inherent measurement inaccuracy, discrepancies between measurement techniques, and individual differences in body types, comparing body fat numbers with your friends is probably not very meaningful.
Keeping the caveats above in mind, it is informative to consider how body fat percentage varies among the population. The United States Centers for Disease Control (CDC) provides the following graph  showing mean body fat percentages from the period 1999-2004, broken down by age group and sex. It is apparent that the average US female ranges between about 32% and 42% body fat during her lifetime, while the average male falls in the range 22% to 32%.
Unfortunately, it should go without saying that nobody should have the goal of attaining the body fat percentage of a typical North American, most of whom (as of 2013-2014) are obese or overweight.
Given that the population average is a poor target, what is a healthy body fat percentage? Very little data exists from which to make concrete conclusions. The American Council on Exercise provides guidelines for body fat percentage ranges. They state  (without details) that "athletes" generally fall in the range 14-20% for women and 6-13% for men. They also list a "fitness" category which shows women in the range 21-24% and men in the range 14-17%. It is difficult to judge the meaning and significance of these numbers without further details.
How to measure body fat
Many methods are available to provide estimates of body fat percentage. Generally speaking, the easier, cheaper, or more convenient the technique, the less accurate and less consistent the results are.
Traditionally, the gold standard measurement for body fat percentage has been based upon a body density analysis. Since fat is substantially less dense than most other body tissues, and the densities of fat and other tissues are approximately known, by measuring the density of the entire body, one can infer the percentage of body fat. Since density is mass/volume, the volume of the body must be determined. This can be done by total immersion in a water tank (displaced water = volume), or in a sealed air chamber by measuring air displacement at different pressures. Unfortunately, water tanks, air chambers, and associated pumps and instrumentation are bulky and expensive, so this is probably not a do-it-yourself option. However many cities have providers with either mobile or in-office equipment who can provide this measurement as a service. For example in the San Diego area, BFIT Body Fat Immersion Testing provides a reasonably priced mobile service.
A newer, more convenient, but still not cheap option is the DXA (Dual X-ray Absorptiometry, sometimes called DEXA) scan. A DXA machine sends X-rays through the body at two different energies. The differences in X-ray absorption at each energy between fat and other tissue can be analyzed to estimate a total body fat percentage. This method gives consistent results which agree well with body density analysis. Although X-rays may sound scary, the dosage is very low. Probably not something you would want to do every day, but not a significant risk. DXA scans can provide a lot of information beyond a single total body fat percentage number, including fat and lean tissue mass in different body regions, bone mineral density, and the degree of asymmetry of muscle and fat distribution. If you live near a major metropolitan area, chances are that a quick Google search will yield a local provider of DXA scans for body fat analysis. In the area near StrengthMetrics, these go for $50-100 and take maybe 15 minutes of your time. We have personally used and can recommend BodyComp, who provide a convenient mobile service for gyms and businesses in the area.
After DXA scans, there are a bunch of cheaper, less accurate methods. Perhaps the most popular are machines based upon bioelectrical impedance. The idea is that since fat and other tissues exhibit different electrical impedances (resistances to flow of electric current), one may be able to infer total body fat percentage by measuring impedance in the body. There are a number of practical and theoretical issues that make this quite a difficult problem, too many to detail in this post. Despite the difficulties, many, many devices are on the market that attempt to estimate body fat percentage using this method. Often these devices are built in to bathroom scales, with electrodes designed to contact the bare feet. It is such a cheap and simple approach in principle that it seems irresistible to manufacturers. Unfortunately, we are not aware of any data establishing the accuracy or consistency of any of these devices. Let the buyer beware.
Another popular method, maybe the most appropriate to apply regularly at home, is skin fold measurement. Essentially, the skin is lightly pinched at various sites on the body, and the thickness of the skin fold is measured with calipers (which are cheap and ubiquitous on Amazon). More body fat correlates with thicker skin folds. Having obtained a number of skin fold thicknesses, various formulae can be applied to generate an estimate of body fat percentage. Unfortunately, that last step is problematic and subject to substantial errors and biases. The good news is that the skin fold thickness measurements themselves are useful as indicators for tracking individual body fat composition, perhaps more useful than the body fat percentage estimate derived from them. For this purpose, it is important to be consistent with the measurement sites and to use the same calipers with the same technique every time.
Hopefully this brief overview is helpful. For the casual fitness enthusiast, a reasonable protocol may be to get DXA scans every six to twelve months, and to measure and record skin fold thicknesses once or twice per month. Future blog posts may cover some of the topics above in more detail.
1. QuickStats: Mean Percentage Body Fat, by Age Group and Sex -- National Health and Nutrition Examination Survey, United States 1999-2004.
2. American Council on Exercise (ACE). Ask the Expert Blog. December 2, 2009.
3. NIH Overweight and Obesity Statistics.