We investigate the degree to which biases in an atmospheric GCM affect the simulation of soil moisture memory within two land surface models (LSMs). The NASA Seasonal-to-Interannual Project (NSIPP) Catchment LSM and the Mosaic LSM were both forced globally offline with a 0.5-degree, 6-hourly, 15-yr (1979-93) dataset constructed from ECMWF reanalysis data that were corrected with observation-based precipitation and radiation. The two simulations were then repeated after imposing GCM climate biases on the forcings -- the forcings were scaled so that their mean seasonal cycles matched those simulated by the NSIPP-1 atmospheric GCM over the same period (1979-93). Four of these supplemental offline global simulations were carried out with each LSM to determine the impact of (1) biased precipitation, (2) biased downward shortwave radiation (3) biased downward long wave radiation and (4) biases in all three forcings together on simulated soil moisture memory. The impact on the memory was found to be significant. The potential for correcting these memory biases through manipulation of soil water holding capacity was explored.