In general, in order to conduct accurate experiments, we need a well-defined exposure field, controllable, ambient conditions (temperature and humidity), and the ability to determine, preferably measure, the SAR and/or temperature of the specimen [1].

Various testing equipment has been developed. For example, free field-anechoic chambers are used to simulate far-field, free-space exposure. They are made of a special material that limits reflections over the frequency range of interest. The chambers are in the form of pyramids or wedges, and the animal is placed inside. Far-field power density by the antenna is given by W = PG/4(pi)r², where r is the distance from the antenna, P is the power delivered to the antenna, and G is the antenna gain [1].

There are some very basic differences between mice and humans, which makes it difficult to make direct comparisons between biological effects on the two species. For instance, humans dissipate heat through sweating, whereas mice cannot. The thermal neutral zone for humans is 24-31°C (75.2-87.8°F), whereas for mice it is 30-33°C (86-91.4°F). Likewise, maximum critical air temp temperatures are different, with 37°C (98.6°F) for mice and 32°C (89.6°F) for an adult human[5]. As seen above (link to dependence of SAR on frequency and shape etc.), body shape and orientation are also very important in determining radiation from radiofrequency fields. Practical methods to account for these differences in models are not available, so caution must be taken.

The basis for this extrapolation is that evolutionary data shows similarities in animals and humans, implying that it is possible that bioeffects may also be similar. But it is important to realize that differences in species mean that effects are not identical from species to another. For this reason, we must be careful in drawing conclusions about implications for humans from animal studies. Ultimately, tests on human subjects need to be performed in order to obtain definitive results.

It is also difficult to measure thermoregulation responses unless animals are restrained, and this adds stress that can interfere with normal thermal regulation and add uncertainty to results. In addition, most studies done have used to 2450MHz radiation, meaning that there is much more to be done for the cellular and PCS systems.

Epidemiological research and in vitro studies are useful, but great progress is also being made in advanced computer and mathematical models, which in addition to experimental data, are proving extremely useful in predicting effects in humans.

In general a physical scaling principle is used to determine the frequency F at which a roughly similar SAR distribution for a man of height L occurs as for an animal of height l exposed at a frequency f (both having the same orientation to the field). The formula is

F = f (l / L).

The SAR distribution will be similar but not identical [1].