Perturbation theory has been used to find the first‐order internal electric field, the SAR (specific absorption rate), the spatial variations of the SAR, and the maximum SAR in prolate spheroidal and ellipsoidal models of man and experimental animals during irradiation by an electromagnetic plane wave when the wavelength is long as compared to the dimensions of the exposed body. In our “conductor” model of man, conductivity is written explicitly in the curlHequation as: ∇ ×H= σE+jω∈E. In what we call the “dielectric” model, the conductivity is contained implicitly in the complex permittivity, so that the curlHequation is ∇ ×H=jω∈E. The two models give different results for first‐order fields because the equations are expanded in a power series ink(k= ω√μ∈); in the conductor model σ enters into the zero‐order equations but in the dielectric model it does not. Because of the nature of the zero‐order equations, the expressions obtained from the conductor model are not valid as σ→0. We have found that the conductor model is valid only if ∈2»ϵ1where ϵ1and ∈2, respectively, are the real and imaginary parts of the complex dielectric constant of the models. Consequently, some caution must be exercised in applying the results of perturbation theory as based on the conductor model. In this paper, the results of perturbation theory as applied to a lossy dielectric ellipsoidal model are described. The SAR in a dielectric ellipsoidal model of a rhesus monkey is calculated and compared with that of the conductor model. The SAR in the two models is found to be the same if the conduction current in the body is much larger than the displacement current. Although the conductor model is inaccurate for low values of conductivity, the equations are simpler than the ones for the dielectric model, and hence t