ABSTRACT: Quantitative fatty acid (FA) signature analysis (QFASA) has recently been developed to estimate the species composition of predator diets by statistically comparing FA signatures of predator adipose tissue with that of their potential prey. Captive feeding trials were used to further test the technique with newly weaned harbour seals Phoca vitulina richardsi (N = 21). Two groups of seals were fed monotypic diets of either Pacific herring Clupea pallasii or surf smelt Hypomesus pretiosus for 42 d while a third group was fed smelt for 21 d followed by herring for 21 d. Blubber biopsies were taken dorsally at Days 0, 21 and 42. Specific calibration coefficients (CC) used by QFASA were developed from 4 juvenile harbour seals and in some cases differed by 2-fold with previously reported phocid CC values. The QFASA diet estimates were evaluated using 2 CC sets, 15 FA subsets and a library of 3 to 11 potential prey species. Diet switches were best tracked using the harbour seal CC and the new FA subset. Overall prey misclassifications were apparent (mean = 12%, range = 4 to 25%) when modeled with 8 additional prey not fed, a trend consistent with overlapping prey FA signatures. Blubber FA turnover rates were not strictly linear and in the order of 1.5 to 3 mo in newly weaned seals. Following parameter optimization of the model, QFASA estimates reflected major diet trends in the feeding study, but were sensitive to the CC and FA subsets used as well as to prey species with similar FA signatures. Our results have important implications in the application of QFASA to the study of pinniped diets with more complex feeding histories and wider prey fields.