Methods and systems for evaluating the quality of a location-determination algorithm of a mobile device are described. An example method may involve receiving a log of sensor data that may include sensor values output by given sensors of a mobile device over a time period, and at least one location estimate for at least one respective point in time within the time period. One or more processors may then determine, using the sensor values, an estimated trajectory that includes a plurality of computed ground-truth locations of the mobile device over the time period. Further, the method may involve determining a difference between a given location estimate and a computed ground-truth location of the plurality of computed ground-truth locations. And the method may involve providing an output indicative of whether the determined difference satisfies a predetermined threshold.