The adequacy of assumptions required to project atmospheric CO2 concentrations in time frames of practical importance is reviewed. Relevant issues concern the form assumed for future fossil fuel release, carbon cycle approximations, and the implications of revisions in fossil fuel patterns required to maintain atmospheric CO2 levels below a chosen threshold. In general, we find that with a judiciously selected exponential fossil fuel release rate, and with a constant airborn fraction, we can estimate atmospheric CO2 growth over the next 50 years based on essentially surprise free scenarios. Resource depletion effects must be included for projections beyond about 50 years, and on this time frame the constant airborne fraction approximation has to be questioned as well (especially in later years when the fossil fuel use begins to taper off). For projections for over 100 years, both energy demand scenarios and currently available carbon cycle models have sufficient uncertainties that atmospheric CO2 levels derived from them are not much better than guesses.