Compared to the common monostatic synthetic aperture radar (SAR), the bistatic SAR data are more challenging to process because of their complicated flight geometry. Although several bistatic SAR image formation algorithms have been presented, most of them are only able to handle azimuth-invariant bistatic configuration where the transmitter and receiver platforms are moving in parallel tracks with identical velocities. This paper proposes an Omega-K algorithm to handle a general case of bistatic SAR data. The two-dimensional frequency spectrum expression derived by the method of series reversion is rewritten to a new form with explicit geometric significance. Based on this new frequency spectrum expression, the frequency mapping function can be derived by making use of the multivariable Taylor's theorem. After data mapping operation, the coupling between the range and azimuth can be eliminated, and the final focused image can be obtained using FFT technology. Numerical simulations are performed to validate the proposed algorithm.