Several UV and near-infrared color selection methods have identified galaxies at z = 1-3. Since each method suffers from selection biases, we have simultaneously applied three leading techniques (Lyman break, BX/BM, and BzK selection) in the Subaru Deep Field. This field has reliable (Δz/(1 + z) = 0.02-0.09) photometric redshifts for 53,000 galaxies from 20 bands (1500 -2.2 μm). The BzK, LBG, and BX/BM samples suffer contamination from z < 1 interlopers of 6%, 8%, and 20%, respectively. Around the redshifts where it is most sensitive (z 1.9 for star-forming BzK, z 1.8 for z 2 LBGs, z 1.6 for BM, and z 2.3 for BX), each technique finds 60%-80% of the census of the three methods. In addition, each of the color techniques shares 75%-96% of its galaxies with another method, which is consistent with previous studies that adopt identical criteria on magnitudes and colors. Combining the three samples gives a comprehensive census that includes 90% of z phot = 1-3 galaxies, using standard magnitude limits similar to previous studies. In fact, we find that among z = 1-2.5 galaxies in the color selection census, 81%-90% of them can be selected by just combining the BzK selection with one of the UV techniques (z 2 LBG or BX and BM). The average galaxy stellar mass, reddening, and star formation rates (SFRs) all decrease systematically from the sBzK population to the LBGs, and to the BX/BMs. The combined color selections yield a total cosmic SFR density of 0.18 0.03 M yr-1 Mpc-3 for K AB ≲ 24. We find that 65% of the star formation is in galaxies with E(B - V) > 0.25mag, even though they are only one-fourth of the census by number. © 2011. The American Astronomical Society. All rights reserved..