In this paper, we analyze the approximate controllability in quadratic mean of some systems governed by stochastic partial differential equations of the Stokes kind. When the noise is state-independent, we obtain satisfactory results, similar to those known for the corresponding deterministic system. In the more complicate case of a multiplicative noise, we are able to give (only) partial results. More precisely, we prove in this case that approximate controllability is equivalent to the unique continuation property for a particular backward (adjoint) stochastic system