tract Radiotelemetry and pellet-group surveys are methods used commonly to measure habi- tat use by large ungulates. However, telemetry can be expensive and analysis of data col- lected from pellet-group surveys is restricted to rank analysis. We explored the feasibil- ity of recording the presence of Roosevelt elk (Cervus elaphus roosevelti) sign to identify habitats used by elk. We surveyed stations (1-ha circular plots) about 0.72 km apart for the presence of 0- to 4-day-old elk sign (tracks and feces) from October to April 1994-1997 at 2 sites in northwestern California. Our objectives were to: 1) measure errors in detecting and classifying elk presence at stations from sign, 2) determine auto- correlation of elk sign at stations to assess what is an independent data point, 3) examine the effect of 2 station sizes on the rate of sign detections, and 4) determine sample sizes needed to detect habitat use. We detected elk sign 96.6% of the time (n=68) when elk were observed at stations within 0-4 days. Elk sign was misclassified only 3 times (n= 70). No autocorrelations in sign detections across time or space were detected because observed data were similar to sign generated randomly at stations. The proportion of 1-ha (0.12) and 2-ha stations (0.13) with sign was similar. Sample sizes >400 were need- ed to have power >0.8 to detect relationships among habitat variables and frequency of sign at stations. Recording the presence of sign in stations appears to be a reliable and feasible technique to measure habitats used by elk.