In analyses using reconstructed $\tau$ leptons, estimation of backgrounds arising for jets misidentified as hadronically decaying $\tau$s becomes a crucial issue.
This paper presents two methods employed by the ATLAS experiment at the LHC, the fully data-driven fake factor method and the semi-data-driven fake rate method. Example of fake factor method in the background modelling of $H^{\pm}\rightarrow\tau\nu$ analysis and the fake rate method applied to the search for high mass resonances decaying to $\tau\tau$, in the $\tau_{ \rm had}\tau_{\rm had}$ channel, are presented. The systematics
associated with the methods are also discussed.