Recently laser cooling of semiconductors has received renewed attention, with the hope that a semiconductor cooler might be able to achieve cryogenic temperatures. In order to study semiconductor laser cooling at cryogenic temperatures, it is crucial that the theory include both the effects of excitons and the electron-hole plasma. In this dissertation, I present a theoreticalanalysis of laser cooling of bulk GaAs based on a microscopic many-particle theory of absorptionand luminescence of a partially ionized electron-hole plasma.This theory has been analyzed from a temperature 10K to 500K. It is shown that at high temperatures (above 300K), cooling can be modeled using older models with a few parameter changes. Below 200K, band filling effects dominate over Auger recombination. Below 30K excitonic effects are essential for laser cooling. In all cases, excitonic effects make cooling easier then predicted by a free carrier model.The initial cooling model is based on the assumption of a homogeneous undoped semiconductor. This model has been systematically modified to include effects that are present in real laser cooling experiments. The following modifications have been performed. 1) Propagation and polariton effects have been included. 2) The effect of p-doping has been included. (n-doping can be modeled in a similar fashion.) 3) In experiments, a passivation layer is required to minimize non-radiative recombination. The passivation results in a npn heterostructure. The effect of the npn heterostructure on cooling has been analyzed. 4) The effect of a Gaussian pump beam was analyzed and 5) Some of the parameters in the cooling model have a large uncertainty. The effect of modifying these parameters has been analyzed.Most of the extensions to the original theory have only had a modest effect on the overall results. However we find that the current passivation technique may not be sufficient to allow cooling. The passivation technique currently used appears to be very good at low densities, but loses some of it's effectiveness at the moderately high densities required for laser cooling. We suggest one possible solution that might enable laser cooling. If the sample can be properly passivated, then we expect laser cooling to be possible.