The solution of a reputational equilibrium is given for a class of linear, quadratic, gaussian dynamic games with noisy control. Although there is imperfect monitoring, a sequential equilibrium is found where the uninformed agents always smoothly learn the type of the informed agent, there is no sudden switch in agents' strategies; a common feature of reputation models. Reputation effects are temporary in the infinite horizon case for positive discount rates, as the discount factor tends to unity there is a permanent reputation.