Uplifted corals in southern Taiwan were dated by the 230Th/234U method using the mass spectrometric techniques. For Holocene samples, our results are in general agreement with earlier studies but with a precision that is a factor of 5-10 better than conventional nuclear counting method and a factor of 2 better than the C-14 dates. Moreover, this technique has the potential of dating samples as young as 30 years and as old as 500,000 years. However, it is also evident that careful selection of samples and thourough field studies are required before the potential of this increased precision can be fully exploited. To investigate the possibility of dating paleoseismic activity, we analyzed four coral samples from near sea surface at Hou-Bi-Hu, Kenting. Three samples yield ages of about 30 to 40 years and one of 1200 years. These results indicate complex coral assemblages and the studies samples may not be suitable for dating paleoseismic activity. A Pleistocene sample from Kuan-Shan at an elevation of 150 meters above sea level yields an age of about 390 kyrs. Although this sample was altered, this date does not support the claim that it was formed at the last interglacial high sea level stand about 120 kyr ago. Two Pleistocene samples from Lung-Keng yields close to equilibrium values, thus implying an age of more than 500 kyrs. This is consistent with the age estimates based on the electron spin resonance method. The occurrence of such an old sample at only 10 meters above the present sea level in Lung-keng which is undergoing rapid uplifting warrants further study.