We tested the reliability of accelerator mass spectrometry (AMS) radiocarbon dates on bone gelatin by dating samples whose maximum or minimum age was constrained by stratigraphic position relative to well-dated volcanic tephra layers The tephra layers were not reworked, and were thick enough to preclude the possibility of redeposition resulting in specimens being found outside the age bounds set by the tephras The damp, fossiliferous, carbonate-rich sediment was in a relatively constant, cool environment in a small cave at nearly 900 m altitude Bones from four species of bird with different diets (a pigeon, a rail, an owlet-nightjar, and a large ratite) and one rodent, the Pacific rat Rattus exulans, were dated The calibrated (calendar) AMS age of each bone was compared to ages predicted from their stratigraphic position and calculated sedimentation rates, inferred from the age of each tephra layer Samples of moa (Aves Dinornithifonnes) eggshell were also dated by AMS, providing an independent dating material None of the bone gelatin or eggshell dates displayed any significant in-built (reservoir) age relative to the marker horizons, nor were any AMS ages significantly younger than expected Our results agree with previous studies in showing that 14 C AMS dates on bone gelatin from deposits that have always had a relatively cool, and stable physical environment are likely to be reliable For bones protected from weathering before burial, and then incorporated in a stable, carbonate-rich environment, the filtered bone gelatin procedure used in this study is adequate to give reliable AMS ages Possible reasons for discrepancies in bone gelatin AMS dates relative to marine shell, or charcoal dates from archaeological sites in dune deposits, are also discussed