Background. Determination of seabird diet usually relies on the analysis of stomach-content remains obtained through stomach flushing; this technique is both invasive and logistically difficult. We evaluate the usefulness of DNA-based faecal analysis in a dietary study on chick-rearing macaroni penguins (Eudyptes chrysolophus) at Heard Island. Conventional stomach-content data was also collected, allowing comparison of the approaches. Methodology/Principal findings. Prey-specific PCR tests were used to detect dietary DNA in faecal samples and amplified prey DNA was cloned and sequenced. Of the 88 faecal samples collected, 39 contained detectable DNA fram one or more of the prey groups targeted with PCR tests. Euphausiid DNA was most commonly detected in the early (guard) stage of chick-rearing, and detection of DNA from the myctophid fish Krefftichthys anderssoni and amphipods became more common in samples collected in the later (crèche) stage. These trends followed those observed in the penguins' stomach contents. In euphauslid-specific clone libraries the proportion of sequences from the two dominant euphausild prey species (Euphausia vallentini and Thysanoessa macrura) changed over the sampling period; again, this reflected the trend in the stomach content data. Analysis of prey sequences in universal clone libraries revealed a higher diversity of fish prey than identified in the stomachs, but non-fish prey were not well represented. Conclusions/Significance. The present study is one of the first to examine the full breadth of a predator's diet using DNA-based faecal analysis. We discuss methodological difficulties encountered and suggest possible refinements. Overall, the ability of the DNA-based approach to detect temporal variation in the diet of macaroni penguins indicates this non-invasive method will be generally useful for monitoring population-level dietary trends in seabirds.