A new global ocean tide model named DTU10 (developed at Technical University of Denmark) representing all major diurnal and semidiurnal tidal constituents is proposed based on an empirical correction to the global tide model FES2004 (Finite Element Solutions), with residual tides determined using the response method. The improvements are achieved by introducing 4 years of TOPEX-Jason 1 interleaved mission into existing 18 years (1993-2010) of primary joint TOPEX, Jason 1, and Jason 2 mission time series. Hereby the spatial distribution of observations are doubled and satellite altimetry should be able to recover twice the spatial variations of the tidal signal which is particularly important in shallow waters where the spatial scale of the tidal signal is scaled down. Outside the +/- 66 degrees parallel combined Envisat, GEOSAT Follow-On, and ERS-2, data sets have been included to solve for the tides up to the +/- 82 degrees parallel. A new approach to removing the annual sea level variations prior to estimating the residual tides significantly improved tidal determination of diurnal constituents from the Sun-synchronous satellites (e. g., ERS-2 and Envisat) in the polar seas. Extensive evaluations with six tide gauge sets show that the new tide model fits the tide gauge measurements favorably to other state of the art global ocean tide models in both the deep and shallow waters, especially in the Arctic Ocean and the Southern Ocean. One example is a comparison with 207 tide gauge data in the East Asian marginal seas where the root-mean-square agreement improved by 35.12%, 22.61%, 27.07%, and 22.65% (M-2, S-2, K-1, and O-1) for the DTU10 tide model compared with the FES2004 tide model. A similar comparison in the Arctic Ocean with 151 gauge data improved by 9.93%, 0.34%, 7.46%, and 9.52% for the M-2, S-2, K-1, and O-1 constituents, respectively.