The goal of this project was to develop new mathematical and computational techniques for quantifying the errors in seismic event locations, focusing on the effects of errors in travel-time predictions from a velocity model. Our approach associates these model errors with the uncertainty in path travel-time corrections inferred from a calibration analysis. The main accomplishment of the project was a general formulation of location uncertainty in terms of the joint inverse problem that combines event location and seismic calibration. The formulation accommodates travel-time nonlinearity, Gaussian and non-Gaussian observational error, and a broad class of parameterizations of path travel-time corrections. We implemented the formulation for the simple correction parameterization as station time terms, applying it to data from the Nevada Test Site (NTS) as a proof of concept of the joint inversion approach. To address the computational intensity of the approach, we re-formulated it as an approximate, two-stage process - calibration followed by location - which we also implemented for the time-term case using newly designed algorithms for each stage. Application of the two-stage approach to NTS data demonstrated its much greater efficiency and suggest it as a feasible uncertainty paradigm for more complex problems in which travel-time corrections are parameterized with 3-D Earth models.