Many statistical analyses of the cosmic velocity held are based on the galaxy number-weighted sampling and therefore cannot escape from the bias effect. We study this problem in the framework of the stochastic bias mechanism using a simplified model and the results of a recent hydrodynamical simulation. We adopt the velocity correlation functions as representative examples and show that velocity statistics might be qualitatively affected by the poorly understood bias mechanism and strongly depend on the tracers of the cosmic velocity field.