To study the effect of ultrafiltration on the radiocarbon ages of relatively poorly preserved bones in Japan, we analyzed the 14C dates of high-molecular-weight (HMW) gelatin samples and compared them with those of other extracted organic fractions, unfiltered gelatin samples extracted from NaOH-treated or NaOH-untreated collagen, and XAD-purified hydrolysates of animal fossil bones (∼4600 BP; gelatin yield of 2–4%) from the Awazu underwater archaeological site, Shiga, Japan. NaOH-treated, unfiltered gelatins and XAD-purified hydrolysates showed statistically similar 14C ages to those of HMW gelatins. The 14C ages of the HMW gelatins were the oldest and similar to those of wood collected from the same layer as the bones, and the NaOH-treated, unfiltered gelatins gave 14C ages within the acceptable margins of error; therefore, ultrafiltration was effective for accurate 14C dating, while NaOH-treated gelatin without ultrafiltration was also sufficient to obtain accurate 14C dates on the animal bones. The 14C ages of human skeletons (∼750 BP; gelatin yield of 2–11%) from 5 individuals excavated from an archaeological site in Yuigahama, Kamakura, Japan, showed statistically the same 14C ages as NaOH-treated, unfiltered gelatins and HMW gelatins within the margins of error, although HMW gelatins were likely to give slightly older ages than unfiltered gelatin with a yield of less than ∼3%. These results indicate that unfiltered gelatins extracted from fossil bones of gelatin yield more than ∼3% can produce accurate 14C ages without the need for ultrafiltration. Ten bone fragments from 3 humans showed the same 14C ages for each individual, suggesting that any bone part from an individual can be used to obtain a representative age. The 14C ages of tooth enamels of 2 individuals were 35 and 70 yr older than their bone ages. Death dates obtained from these age gaps agreed with those determined by morphology.