During the last deglaciation, the ventilation of the subarctic Pacific is hypothesized to have changed dramatically, including the rejuvenation of a poorly ventilated abyssal water mass that filled the deep ocean, and fluctuations in the strength of North Pacific intermediate and deep water formation at millennial timescales. Foraminiferal radiocarbon reconstructions of past ventilation changes in the Pacific are valuable but are hampered by poor carbonate preservation, low sediment accumulation rates, bias from bioturbation, and poorly constrained past surface reservoir age. In this study, we present paired benthic-planktonic radiocarbon measurements from the Okhotsk Sea and Emperor Seamounts. We take advantage of large contemporaneous peaks in benthic abundances from the last glacial maximum, Bolling-Allerod (BA), and early Holocene to produce time slices of radiocarbon from 1 to 4 km water depth. We explore the impact of uncertain surface reservoir age and evaluate several approaches to quantifying past ocean radiocarbon distribution using our NW Pacific data and a compilation of published data from the North Pacific. Both the calendar age and the absolute value of an ocean radiocarbon estimate depend on the assumed surface reservoir age. But for a time slice from a small geographical area with radiocarbon-independent stratigraphic correlation between cores, the shape of a water column profile is independent of surface reservoir age. The NW Pacific profiles are similar in shape to the compilation profiles for the entire North Pacific, which suggests that deglacial surface reservoir age changes across the N Pacific did not diverge dramatically across the areas sampled. The Last Glacial Maximum (LGM) profile >2 km spans a wide range of values, ranging from values similar to today to lower than today. However, by the BA the profile has a similar shape to today. Ultimately, local surface reservoir ages, end-member water mass composition, and mixing ratios must each be constrained before a radiocarbon activity reconstruction can be used to confidently infer ventilation changes.