We used 10 Telonics ST-3 platform transmitter terminals (PTT's) configured for wolves and ungulates to examine the performance of the Argos satellite telemetry system. Under near-optimal conditions, 68 percentile errors for location qualities (NQ) 1, 2, and 3 were 1,188, 903, and 361 m, respectively. Errors (r(E)) exceeded expected values for NQ = 2 and 3, varied greatly among PTT's, increased as the difference (H(E)) between the estimated and actual PTT elevations increased, and were correlated nonlinearly with maximum satellite pass height (P(H)). We present a model of the relationships among r(E), H(E), and P(H). Errors were bimodally distributed along the east-west axis and tended to occur away from the satellite when H(E) was positive. A southeasterly bias increased with H(E), probably due to the particular distribution of satellite passes and effects of H(E) on r(E). Under near-optimal conditions, greater-than-or-equal-to 1 sensor message was received for up to 64% of available (P(H) greater-than-or-equal-to 5-degrees) satellite passes, and a location (NQ greater-than-or-equal-to 1) was calculated for up to 63% of such passes. Sampling frequencies of sensor and location data declined 13 and 70%, respectively, for PTT's in a valley bottom and 65 and 86%, respectively, for PTT's on animals that were in valley bottoms. Sampling frequencies were greater for ungulate than for wolf collars.
