This thesis examines the effects of time shifts in density models and their effects on orbit prediction. Empirical density models are subject to lags in their prediction of atmospheric density, especially during times of high geomagnetic activity. An analytical density model is used to demonstrate that time delays can cause errors in the satellite orbit, and also that the errors can increase as satellite height decreases. The JB2008 and NRLMSISE-00 models are examined here. The models are first compared to densities that were derived from accelerometers on the CHAMP spacecraft. Satellite orbits are integrated using each of the models with the best available inputs and the CHAMP density. Errors resulting from the models are seen to reach up to several kilometers, with the JB2008 model performing the best. Time shifts are then introduced to the models, and they are each compared to a model with the best available inputs. The time shifts range from 1 to 6 hours. It is shown that errors increase for larger shifts, up to several kilometers again, with the NRLMSISE-00 model performing the best for the shifted densities. Finally shifts in real world density fluctations are examined by smoothing the CHAMP densities to remove short term orbital variations, and then shifting these densities by 1 to 6 hours. The errors shown in this case again reach several kilometers. The errors seen are shown to be significant to various spacecraft operations for all cases. The largest errors in all cases are also shown to occur during times of high geomagnetic activity.