This chapter presents a study of ocean flow field reconstruction using profiling floats. Buoyancy-controlled, semi-autonomous profiling floats provide an economic and flexible means to monitor ocean flows. However, a key challenge in enabling this capability lies in how to extract the velocities of the ocean flow at sampled locations from the recorded motion data of floats. This leads to a problem that can be cast as the joint estimation of a nonlinear dynamic system’s inputs and states. This chapter develops solutions from the perspective of Bayesian estimation and evaluates their application to ocean flow field reconstruction through simulations. The proposed results may also be useful in many other scientific and engineering problems involving input and state estimation.