This paper studies how propagation and termination losses affect full transient-based fault location techniques. Their accuracy is discussed in terms of both location uncertainty, caused by a limited spatial resolution, and systematic errors, caused by a bias in the fault-location metrics. This last case is proven to be by far likelier when propagation losses are higher than the dissipation in line termination loads. Two different location metrics are studied, namely correlation and normalized projection, as found in the literature, with correlation proven to be unbiased, since it benefits from two location mechanisms, namely frequency and time-decay matching of a line resonances, as opposed to projection, which only relies on the former mechanism. A numerical analysis of realistic lossy overhead lines confirms theoretical predictions about biased fault location and loss of spatial resolution and the role played by the frequency content of transient data. When applied to the modal analysis of a three-phase transmission line, these results help explaining why faults are located with widely variable accuracy depending on their distance and the bandwidth of the recorded transient, confirming that wide-band transient sampling does not necessarily results in the best location accuracy.