Orthogonal frequency division multiplexing (OFDM) is a popular modulation technique that can combat impulsive noise, is robust to multipath fading, is spectrally efficient, and can allow flexible allocation of spectrum. It has become a key standard in cognitive radio systems as well as an enabling technology for mobile data access systems. An OFDM receiver's performance is heavily impacted by the accuracy of its symbol timing offset (STO) and carrier frequency offset (CFO) estimation. This paper proposes a novel OFDM synchronization method that combines robust performance with computational efficiency. FPGA prototyping is used to explore the trade-off between the number of computations to be performed and computation word length with respect to both synchronization performance and power consumption. Through simulation, the proposed method is shown to provide accurate fractional CFO estimation as well as STO estimation in a range of channels. In particular, it can yield excellent synchronization performance in the face of a CFO that is larger than many state-of-the-art synchronization implementations can handle. The system implementation demonstrates efficient resource usage and reduced power consumption compared with existing methods, and this is explored as a fine-grained trade-off between performance and power consumption. The result is a robust method suitable for use in low-power radios, enabling less precise analog front ends to be used. © 2014 Springer Science+Business Media New York.