We perform a statistical weak lensing analysis of dark matter profiles around tracers of halo mass from galactic- to cluster-size halos. In this analysis we use 170 640 isolated ∼ L∗ galaxies split into ellipticals and spirals, 38 236 groups traced by isolated spectroscopic Luminous Red Galaxies (LRGs) and 13 823 MaxBCG clusters from the Sloan Digital Sky Survey (SDSS) covering a wide range of richness. Together these three samples allow a determination of the density profiles of dark
matter halos over three orders of magnitude in mass, from 1012M⊙ to 1015M⊙. The resulting lensing signal is consistent with an NFW or Einasto profile on scales outside the central region. In the inner regions, uncertainty in modeling of the proper identification of the halo center and inclusion of baryonic effects from the central galaxy make the comparison less reliable. We find that the NFW concentration parameter c200b decreases with halo mass, from around 10 for galactic halos to 4 for cluster halos. Assuming its dependence on halo mass in the form of c200b = c0(M/1014h−1M⊙)− we find c0 = 4.6 ± 0.7 (at z = 0.22) and = 0.13 ± 0.07, with very similar results for the Einasto profile. The slope () is in agreement with theoretical predictions, while the amplitude is about two standard deviations below the predictions for this mass and redshift, but we note that the published values in the literature differ at a level of 10-20% and that for a proper comparison our analysis should be repeated in simulations. We compare our results to other recent determinations, some of which find significantly higher concentrations. We discuss the implications of our results for the baryonic effects on the shear power spectrum: since these are expected to increase the halo concentration, the fact that we see no evidence of high concentrations on scales above 20% of the virial radius suggests that baryonic effects are limited to small scales, and are not a significant source of uncertainty for the current weak lensing measurements of the dark matter power spectrum.