Using non-lethal tissue sampling for stable isotope analysis has become standard in many fields, but not for fishes, despite being desirable when species are rare or protected, when repeated sampling of individuals is required or where removal may bias other analyses. Here, we examine the utility of fish dorsal fin membrane as an alternative to muscle for analyzing δ13C and δ15N ratios in two reef fish species (blue cod Parapercis colias and spotty Notolabrus celidotus) that have differing feeding modes. Both species exhibited evidence of size-based feeding from fin δ15N values, but not from muscle. Blue cod fin d15N increased steadily throughout the sampled size range (213–412 mm fork length), whereas spotty exhibited a distinct ontogenetic diet shift at approximately 120–140 mm fork length after which size-based feeding did not occur. Fin membrane was higher than muscle in δ13C in both species and in d15N for blue cod, but fin δ15N was lower than muscle in spotty. The δ13C and δ15N fin–muscle offsets were constant in spotty regardless of size, while in blue cod, δ13C was constant with fish size, but δ15N offsets increased with increasing fish size. Non-lethal sampling utilizing fin tissue can be employed to estimate stable isotope values of muscle in fishes, but it is necessary to assess relationships among tissues and the effects of fish size on isotope values a priori for each species studied. Our data indicated that fin membrane may be a more sensitive tissue than muscle for detecting size-based feeding in some fish species using stable isotopes. A critical literature review revealed inconsistencies in tissue types tested, little understanding of tissue-specific trophic shift or turnover rates, and pseudo-replicated analyses leading to erroneous postulating of 1:1 relationships between tissues.