The monthly-mean total chlorine abundance (ClT) at 55 km inferred from HALOE HCl observations increases from 1992 to 1997 and then subsequently decreases. The pre-1997 increase is consistent with surface measurements of ClT time-lagged by around 6 years. However, a decrease after 1997 is inconsistent with such a time lag, which would predict a peak in late 1999. Accounting for stratospheric mixing processes produces an expected stratosphere ClT which is in agreement with the HALOE ClT time series considering the uncertainty in the HALOE data. However, the peak in ClT is still predicted to occur in later 1999 rather than 1997. We find that reasonable low frequency changes in transport, chlorine partitioning, anomalous buildup of organic chlorine at 55 km, and tropospheric rainout of inorganic chlorine do not reconcile the expected and HALOE ClT time series. At present, we are unable to explain how upper stratospheric ClT could decrease as early as 1997. ¿ 2001 American Geophysical Union