Do the large economic benefits predicted for Olympics host countries actually materialise? We re-examine the Sydney 2000 Olympics via historical modelling with a multiregional dynamic computable general equilibrium model, taking care to avoid sources of overestimation, such as elastic factor supply assumptions; failure to treat public inputs as costs; and overestimation of foreign tourism demand impacts. We first conduct a simulation from 1997/1998 to 2005/2006 to construct a base case and to assess the extent of Olympics-induced awareness of Australia on tourism demand. Our historical simulation results which are driven by observed values for economic variables, including tourism statistics, do not provide support for the presence of induced tourism. We then conduct a simulation for a no-Sydney Olympics counterfactual and find that the Olympics generated a real consumption loss of $2.1 billion. We do identify economic winners from the Games and consider the possibility of non-use benefits from the Games offsetting the real consumption loss.