This study compared diet reconstructed from different compartments of the digestive tract of harp seals (Pagophilus groenlandicus (Erxleben, 1777)) with the diet estimated using stable carbon and stable nitrogen isotope ratios in mixing models. Diet composition in 18 feeding harp seals (mean age = 2.4 years, SD = 1.8 years, range = 0–6 years) was determined using traditional methods of hard-part identification and reconstruction, and stable carbon and stable nitrogen isotope ratios. Diet composition consisted of 68.8% (SD = 8.7%) and 69.6% (SD = 11.6%) by mass of invertebrates or 65.0% (SD = 8.8%) and 66.5% (SD = 11.8%) by energy of invertebrates for the stomach and small-intestine compartments, respectively. Reconstructed diets using material recovered from the large-intestine contents suggested a diet of 43.1% (SD = 12.2%) and 38.0% (SD = 11.9%) invertebrates using mass and energy, respectively. Stable carbon and stable nitrogen isotope ratios determined for the same individual harp seals suggested a diet consisting of approximately 66.1% (SD = 117.4) invertebrates, indicating that diet reconstructions based on hard parts from stomachs are likely to be more representative than reconstructions from large-intestine contents. In species that feed on a combination of vertebrates and invertebrates, the use of faecal material to reconstruct diet composition will likely underestimate the importance of invertebrates in the diet.