Existing indices of satellite telemetry error offer objective standards for censoring poor locations, but have drawbacks. Examining distances and relative directions between consecutive satellite telemetry locations, I developed an alternative error index, xi, and compared its performance with that of the location quality index, NQ (Serv. Argos 1988). In controlled tests, xi was more (P less-than-or-equal-to 0.005) effective for improving precision than was a threshold of NQ > 1. The xi index also conferred greater control over the trade off between sample size and precision, making xi more cost-effective than NQ. Performances of xi and NQ were otherwise comparable. In field tests with bighorn sheep (Ovis canadensis), rejecting locations where xi greater-than-or-equal-to 1.5 km reduced (P < 0.001) longitudinal dispersion, the predominant error component. Longitudinal dispersion for these locations was less (P = 0.025) than for locations where NQ > 1 and 63% fewer data were censored, so that the extent of animals' movements was better indicated by using xi rather than NQ. Because use of xi may lead to underestimating the number of long-range, short-term forays (especially when the frequency of forays is high relative to sampling frequency), potential bias should be considered before using xi. Nonetheless, xi should be a useful alternative to NQ in many animal-tracking studies.
