We compared the effectiveness of three amphibian survey methods (trapping, light-touch, and destructive sampling) in seeps because the efficiency of these methods in this kind of habitat has not been evaluated previously. Our study sites were located on managed forests in southwest Washington State. Trapping involved setting up an array of funnel traps across the seep, light-touch is a visual-encounter method facilitated by overturning and replacing moveable cover objects, and destructive sampling consisted of searching the seep surface by excavating the top 15 cm of soil and dismantling woody debris. Trapping and light-touch were compared through six, three-week periods, whereas the non-repeatable destructive sampling was compared with trapping and light-touch only during the final sampling period. Light-touch detected more species (P = 0.007), more coastal tailed frogs (Ascaphus truei; P < 0.001), and similar numbers of Columbia torrent salamanders (Rhyacotriton kezeri; P = 0.123) and western red-backed salamanders (Plethodon vehiculum; P = 0.152) compared to trapping. When compared to other survey methods during the final sampling period, destructive sampling detected more species (P = 0.001) and more torrent salamanders (P= 0.005) than trapping, but detected similar numbers of species (P = 0.15) and torrent salamanders (P = 0.21) as light-touch. Light-touch was less expensive in material costs and required fewer visits, but more time (77 vs. 19 person-minutes) per survey session than trapping. Destructive sampling had the same material costs as light-touch, but required more time per survey session (690 person-minutes) than either of the other two methods. Where a repeatable method is required, light-touch seems preferable to trapping because it enumerates a higher percentage of species and individuals, has fewer potential survey biases, and can provide data on within-seep amphibian use.