The response of turbulent premixed flames to inlet velocity fluctuations is studied experimentally in a lean premixed, swirl-stabilized, gas turbine combustor. Overall chemiluminescence intensity is used as a measure of the fluctuations in the flame’s global heat release rate, and hot wire anemometry is used to measure the inlet velocity fluctuations. Tests are conducted over a range of mean inlet velocities, equivalence ratios, and velocity fluctuation frequencies, while the normalized inlet velocity fluctuation (V′/Vmean) is fixed at 5% to ensure linear flame response over the employed modulation frequency range. The measurements are used to calculate a flame transfer function relating the velocity fluctuation to the heat release fluctuation as a function of the velocity fluctuation frequency. At low frequency, the gain of the flame transfer function increases with increasing frequency to a peak value greater than 1. As the frequency is further increased, the gain decreases to a minimum value, followed by a second smaller peak. The frequencies at which the gain is minimum and achieves its second peak are found to depend on the convection time scale and the flame’s characteristic length scale. Phase-synchronized CH∗ chemiluminescence imaging is used to characterize the flame’s response to inlet velocity fluctuations. The observed flame response can be explained in terms of the interaction of two flame perturbation mechanisms, one originating at flame-anchoring point and propagating along the flame front and the other from vorticity field generated in the outer shear layer in the annular mixing section. An analysis of the phase-synchronized flame images show that when both perturbations arrive at the flame at the same time (or phase), they constructively interfere, producing the second peak observed in the gain curves. When the perturbations arrive at the flame 180 degrees out-of-phase, they destructively interfere, producing the observed minimum in the gain curve.