An important component of probabilistic risk assessment methods is the development of models to quantify the direct consequences of damage to geo-structural components for a given intensity of the hazard. This paper presents a general probabilistic framework for correlated repair cost and downtime estimation of geo-structures exposed to seismic hazards. The framework uses as input the results of nonlinear time-history analysis of geo-structures for the set of earthquake records that are representative of the seismic hazard models for the region of interest. The repair cost and downtime are estimated for individual earthquakes probabilistically considering the uncertainties associated with damage states. In addition, the formulation of the repair cost and downtime accounts for the reduction in the repair requirements as the number of damaged components in the given damage state increases. An analytical linear and two bilinear regression models are proposed for conditional correlated seismic repair cost and downtime estimation of geo-structures given the intensity measure. The proposed framework is demonstrated by developing seismic repair models of a typical pile-supported wharf structure on the west coast of the United States. The presented framework is general and can be applied to other types of geo-structures and hazards and can include other decision variables such as loss of life as well. Copyright © 2013 John Wiley & Sons, Ltd.