An algorithm is presented for estimating unknown system dynamics and unpredictable disturbances. Estimation is performed by convolving system input/output data with a very simple model of the plant. Dynamics are estimated directly, without estimating model parameters or measuring state derivatives. The estimation algorithm is used in the context of time delay control to provide accurate control of a multi-input, multi-output nonlinear plant in both simulations and experiments. The control action is continuous and accurate upon system startup. The simulation and experimental data show excellent control performance. Computer simulations and experimental data indicate that the proposed method for estimating unknown dynamics is a viable alternative to derivative feedback, previously required for estimation of unknown dynamics in the time delay control algorithm.>