Coral reefs are in decline worldwide from a combination of natural and human forces. The environmental compromises faced by coral reef habitats and their associated fishes are potentially stressful, and in this study we examined the potential for assessing stress levels in coral reef fish. We determined the feasibility of using fecal casts from parrotfishes for remote assessment of stress-related hormones (cortisol and corticosterone), and the response of these hormones to the stress of restraint and hypoxia. Measurement of these hormones in fecal extracts by high performance liquid chromatography (HPLC) was validated using mass spectrometry, chemical derivitization, and radioactive tracer methods. In aquarium-adapted parrotfish, baseline levels of cortisol and corticosterone averaged 3.4??1.1 and 14.8??2.8ng/g feces, respectively, across 32 days. During 13 days of periodic stress these hormones, respectively, average 10.8-fold and 3.2-fold greater than baseline, with a return to near baseline during a 23-day follow-up. Testosterone was also measured as a reference hormone which is not part of the stress-response axis. Levels of this hormone were similar across the study. These fecal hormones were also measured in a field study of parrotfish in 10 fringing coral reef areas around the Caribbean Island of St. John, US Virgin Islands. Extracts of remotely collected fecal casts of three parrotfish species revealed no difference in respective average hormone levels among these species. Also, there was no difference in respective hormone levels between aquarium and field environments. However, levels of both cortisol and corticosterone, but not testosterone, were elevated in two of the 10 reef sites surveyed. This study demonstrates that parrotfish fecals can be collected in aquarium and field conditions and that steroid hormones in these fecals can be extracted and reliably measured. The study also demonstrates that cortisol and corticosterone in parrotfish fecals can be used as an indicator of the stress-response which is unlikely to be masked by intrinsic variability in the sample source, environment or methodology. ?? 2003 Elsevier Science (USA). All rights reserved.