In recent years, the most common technique for radiocarbon dating of deep-ocean sediments has been accelerator mass spectrometry (AMS) analysis of hand-picked planktonic foraminifera (forams). Some studies have exposed age offsets between different sediment size fractions from the same depth within a core and this has important implications when establishing a chronological framework for palaeoceanographic records associated with a particular sediment component. The mechanisms generating the age offsets are not fully understood, a problem compounded by the fact that the fraction defined as “large” varies between different studies. To explore this problem, we dated samples of hand-picked forams from two Biogeochemical Ocean Flux Study (BOFS) cores, for which the presence of an offset between the bulk carbonate and >150 μm foraminiferal calcite had already been demonstrated. The presence of a constant age offset between bulk carbonate and coarse fraction material at the two BOFS sites has been confirmed, but the magnitude of the offset is dependent on whether a simple size-separation technique or hand-picking of well-preserved forams is applied. This may be explained if the selection of well preserved forams biases the sample towards those specimens that have spent least time in the surface mixed layer (SML) or have undergone less size selective mixing. Modeling of the 14C profiles demonstrates that SML depth and sediment accumulation rates are the same for both the bulk and coarse sediment fractions, which is consistent with the hypothesis that size-selective mixing is responsible for the age offset.