The present study investigated how concentrations of different organochlorines (OCs) vary with nutritional condition in adult harp seal (Phoca groenlandica) females in a captive fasting experiment and in the wild. During the first part of this study seals in good condition (N= 5) were fasted for a period of 28 days and blood and blubber samples were collected at days 1, 14 and 28. Blubber OC concentrations remained unchanged throughout the experimental period, and were consistently significantly higher than concentrations in the blood. In contrast to blubber OC levels, blood OC levels showed a significant time-dependant increase. During the fasting experiment the seals lost an average of 24 kg of body mass, which is less than half the natural annual variation in this species in the wild. The second part of this study compared OC concentrations in blood and blubber from seals collected at prime condition before the breeding season (N = 10) with animals collected during molt when condition is poor (N = 7). The average mass difference between the two groups was more than 40 kg. Blood levels of most OCs were significantly higher in the thin seals compared with the levels found in the fat seals. These differences in blood OC concentrations were much greater than what was found during the fasting experiment. For example sigmaPCB levels in the blood during the fasting experiment increased by approximately 83%, while the corresponding increase in blood levels of sigmaPCB between the two other samples was 720% (from 201 to 1,447 ng/g lipid). Blubber levels of OCs from the animals in the wild were significantly higher in the thin seals compared with the fat animals. In addition, the blubber levels of OCs were significantly higher than blood levels of OCs only for the fat seals. This study demonstrates the extreme variability present in the concentrations of OCs in blood of seals in response to change in condition. Since the natural variation in condition is extreme during phocid seals' annual cycles, we recommend that blood should not be used in studies of OCs where the aim of the study is to monitor OC levels for comparative purposes or time-trend analysis.