Natural geochemical signatures in calcified structures are commonly employed to retrospectively estimate dispersal pathways of larval fish and invertebrates. However, the accuracy of the approach is generally untested due to the absence of individuals with known dispersal histories. We used genetic parentage analysis (genotyping) to divide 110 new recruits of the orange clownfish, Amphiprion percula, from Kimbe Island, Papua New Guinea, into two groups: “self-recruiters” spawned by parents on Kimbe Island and “immigrants” that had dispersed from distant reefs (>10km away). Analysis of daily increments in sagittal otoliths found no significant difference in PLDs or otolith growth rates between self-recruiting and immigrant larvae. We also quantified otolith Sr/Ca and Ba/Ca ratios during the larval phase using laser ablation inductively coupled plasma mass spectrometry. Again, we found no significant differences in larval profiles of either element between self-recruits and immigrants. Our results highlight the need for caution when interpreting otolith dispersal histories based on natural geochemical tags in the absence of water chemistry data or known-origin larvae with which to test the discriminatory ability of natural tags.