Environmental DNA (eDNA) methods for detecting and estimating abundance of aquatic species are emerging rapidly, but little is known about how processes such as secretion rate, environmental degradation, and time since colonization or extirpation from a given site affect eDNA measurements. Using stream-dwelling salamanders and quantitative PCR (qPCR) analysis, we conducted three experiments to assess eDNA: (i) production rate; (ii) persistence time under different temperature and light conditions; and (iii) detectability and concentration through time following experimental introduction and removal of salamanders into previously unoccupied streams. We found that 44–50 g individuals held in aquaria produced 77 ng eDNA/h for 2 h, after which production either slowed considerably or began to equilibrate with degradation. eDNA in both full-sun and shaded treatments degraded exponentially to <1% of the original concentration after 3 days. eDNA was no longer detectable in full-sun samples after 8 days, whereas eDNA was detected in 20% of shaded samples after 11 days and 100% of refrigerated control samples after 18 days. When translocated into unoccupied streams, salamanders were detectable after 6 h, but only when densities were relatively high (0.2481 individuals/m2) and when samples were collected within 5 m of the animals. Concentrations of eDNA detected were very low and increased steadily from 6–24 h after introduction, reaching 0.0022 ng/L. Within 1 h of removing salamanders from the stream, eDNA was no longer detectable. These results suggest that eDNA detectability and concentration depend on production rates of individuals, environmental conditions, density of animals, and their residence time.