Trace nonrecycling impurities (scandium and CaF2) have been injected into Alcator C-Mod plasmas in order to determine impurity transport coefficient profiles in a number of operating regimes. Recycling Ar has also been injected to characterize steady-state impurity density profiles. Subsequent impurity emission has been observed with spatially scanning X-ray and vacuum ultraviolet spectrometer systems, in addition to very high spatial resolution X-ray and bolometer arrays viewing the plasma edge. Measured time-resolved brightness profiles of helium-, lithium-, and beryllium-like transitions have been compared with those calculated from a transport code that includes impurity diffusion and convection, in conjunction with an atomic physics package for individual line emission. Similar modeling has been performed for the edge observations, which are unresolved in energy. The line time histories and the profile shapes put large constraints on the impurity diffusion coefficient and convection velocity profiles. In L-mode plasmas, impurity confinement times are short (~20 ms), with diffusivities in the range of 0.5 m2/s, anomalously large compared to neoclassical values. During Enhanced D (EDA) H-modes, the impurity confinement times are longer than in L-mode plasmas, and the modeling suggests that there exists inward convection (50 m/s) near the plasma edge, with greatly reduced diffusion (of order 0.1 m2/s), also in the region of the edge transport barrier. These edge values of the transport coefficients during EDA H-mode are qualitatively similar to the neoclassical values. In edge localized mode-free H-mode discharges, impurity accumulation occurs, dominated by large inward impurity convection in the pedestal region. A scaling of the impurity confinement time with H-factor reveals a very strong exponential dependence. In internal transport barrier discharges, there is significant impurity accumulation inside of the barrier foot, typically at r/a = 0.5. Steady-state impurity density profiles in L-mode plasmas have a large up-down asymmetry near the last closed flux surface. The impurity density enhancement, in the direction opposite to the ion B × [nabla]B drift, is consistent with modeling of neoclassical parallel impurity transport.