Both the Thellier method and Shaw's method were applied to samples of Leg 55 basalts to determine intensities of the geomagnetic field in the early Cenozoic. The success rates in the experiments were 5/16 (Shaw's method) and 8/32 (the Thellier method). Internal consistency of data cannot be demonstrated, because only one paleointensity estimate is available for individual flow units, except Flow Unit 2 of Ojin. Results from Shaw's method are not very good, in that some change occurred in ARM by heating and the ARMs lack the coercivity interval between 800 and 1000 Oe. For many samples, a demagnetizing field higher than 1000 Oe seems necessary to obtain reliable data. In successful results from the Thellier method, linearity of NRM and TRM components occurs approximately between 100 degrees and 500 degrees C. Deviation from the linear line at low temperatures is a result of a superposed VRM component in the present field direction. Deviation at high temperatures is probably caused by some change in chemistry or grain-size distribution of ferromagnetic minerals. This point is discussed in a separate paper (Kono, this volume). The mean and standard deviation of VDMs calculated from the paleointensity results presented here are 13.1 + or -9.2 x 10 (super 25) g cm (super 3) . The mean is significantly larger and the range much wider than the corresponding values for the Deccan basalts of similar age (7.2 + or -3.0 x 10 (super 25) g cm (super 3) ). In Hole 433C (Suiko), flow units with inclinations much different from the hole mean inclination tend to give very large paleointensities. To determine whether this indicates the time when the non-dipole component was very large or the time of geomagnetic reversals or excursions is left for future studies.