The bacterial flagellar motor and the ATP–hydrolysing F 1 portion of the F 1 F o –ATPase are known to be rotary motors, and it seems highly probable that the H + –translocating F o portion rotates too. The energy source in the case of F o and the flagellar motor is the flow of ions, either H + (protons) or Na + , down an electrochemical gradient across a membrane. The fact that ions flow in a particular direction through a well–defined structure in these motors invites the possibility of a type of mechanism based on geometric constraints between the rotor position and the paths of ions flowing through the motor. The two beststudied examples of such a mechanism are the ‘turnstile’ model of Khan and Berg and the ‘proton turbine’ model of Lauger or Berry. Models such as these are typically represented by a small number of kinetic states and certain allowed transitions between them. This allows the calculation of predictions of motor behaviour and establishes a dialogue between models and experimental results. In the near future structural data and observations of single–molecule events should help to determine the nature of the mechanism of rotary motors, while motor models must be developed that can adequately explain the measured relationships between torque and speed in the flagellar motor.