We investigate satellite orbital drag effects at low-Earth orbit (LEO) associated with thermosphere heating during magnetic storms caused by coronal mass ejections. CHAllenge Mini-satellite Payload (CHAMP) and Gravity Recovery And Climate Experiment (GRACE) neutral density data are used to compute orbital drag. Storm-to-quiet density comparisons are performed with background densities obtained by the Jacchia-Bowman 2008 (JB2008) empirical model. Our storms are grouped in different categories regarding their intensities as indicated by minimum values of the SYM-H index. We then perform superposed epoch analyses with storm main phase onset as zero epoch time. In general, we find that orbital drag effects are larger for CHAMP (lower altitudes) in comparison to GRACE (higher altitudes). Results show that storm-time drag effects manifest first at high latitudes, but for extreme storms particularly observed by GRACE stronger orbital drag effects occur during early main phase at low/equatorial latitudes, probably due to heating propagation from high latitudes. We find that storm-time orbital decay along the satellites' path generally increases with storm intensity, being stronger and faster for the most extreme events. For these events, orbital drag effects decrease faster probably due to elevated cooling effects caused by nitric oxide, which introduce modeled density uncertainties during storm recovery phase. Errors associated with total orbit decay introduced by JB2008 are generally the largest for the strongest storms, and increase during storm times, particular during recovery phases. We discuss the implication of these uncertainties for the prediction of collision between space objects at LEO during magnetic storms.