Intracellular recordings were made from turtle cochlear hair cells in order to study the changes in their tuning properties resulting from electrical stimulation of the efferent axons. Efferent stimulation caused a reduction in the amplitude of the receptor potential at the hair cell's most sensitive or characteristic frequency, an increased amplitude at frequencies more than an octave below the characteristic frequency, and no change at very high frequencies. These differential effects resulted in a broadening of each cell's tuning curve, which, during maximal efferent stimulation degenerated from a sharply tuned resonance to a critically damped low-pass filter. Efferent alterations in tuning were also inferred from the oscillations in membrane potential produced by acoustic clicks or extrinsic currents. The quality factor (Q) of tuning, derived from the decay of the oscillations, was progressively reduced with synaptic hyperpolarizations up to about 5 mV in amplitude. A consequence of efferent action was that the wave forms of transient pressure changes were more faithfully encoded as changes in hair cell membrane potential. Hyperpolarization of a hair cell by steady current injection resulted in a lowering of its characteristic frequency and quality factor, and an increase in steady-state resistance. By comparison, for a given reduction in quality factor, efferent stimulation was associated with a smaller change in characteristic frequency. This difference is expected if the resonance is also damped by the shunting action of the synaptic conductance. Perfusion with perilymphs containing 0.5-15 mM of the potassium channel blocker, tetraethylammonium bromide (TEA) reduced the hair cell's frequency selectivity, whether assayed acoustically or with extrinsic currents. Lower TEA concentrations abolished the efferent inhibitory post-synaptic potential with only a minor change in tuning. TEA produced other effects different from efferent stimulation including (i) a lowering of the characteristic frequency, and (ii) a highly asymmetric receptor potential. These observations suggest that the efferents do not simply block membrane conductances associated with tuning. We conclude that the efferent modification of the shape of the tuning curve may be a composite result of the synaptic conductance and the hyperpolarization of the hair cell membrane.