In general, CO2 sequestration by carbonation is estimated by laboratory experimentation and geochemical simulation. In this study, however, estimation is based on a natural analogue study of the Miocene basalt in the Kuanhsi-Chutung area, Northwestern Taiwan. This region has great potential in terms of geological and geochemical environments for CO2 sequestration. Outcropping Miocene basalt in the study area shows extensive serpentinization and carbonation. The carbon stable isotopes of carbonates lie on the depleted side of the Lohmann meteoric calcite line, which demonstrates that the carbonates most probably precipitate directly from meteoric fluid, and water-rock interaction is less involved in the carbonation process. Oxygen stable isotope examinations also show much depleted ratios, representative of product formation under low temperatures (∼50-90°C). This translates to a depth of 1-2km, which is a practical depth for a CO2 sequestration reservoir. According to petrographic observation and electron microprobe analysis, the diopside grains in the basalt are resistant to serpentinization and carbonation; therefore, the fluid causing alteration is likely enriched with calcium and there must be additional sources of calcium for carbon mineralization. These derived geochemical properties of the fluid support the late Miocene sandstone and enclosed basalts as having high potential for being a CO2 sequestration reservoir. Moreover, the existing geochemical environments allow for mineralogical assemblages of ultramafic xenoliths, indicating that forsterite, orthopyroxene and feldspar minerals are readily replaced by carbonates. Based on the mineral transformation in xenoliths, the capacity of CO2 mineral sequestration of the Miocene basalt is semi-quantitatively estimated at 94.15kg CO2 chemically trapped per 1m3 basalt. With this value, total CO2 sequestration capacity can be evaluated by a geophysical survey of the amount of viable Miocene basalt at the potential sites. Such a survey is required in the near future.