In-phase/quadrature (I/Q) imbalances are front-end impairments which may greatly complicate the synchronization task in a low-cost direct-conversion receiver (DCR). In this paper we investigate the possibility of using the space-alternating generalized expectation-maximization (SAGE) algorithm to recover the carrier frequency offset in OFDM terminals with a DCR architecture. Our study leads to a novel scheme that operates in a recursive fashion and exploits a conventional OFDM training preamble composed by several repeated parts. At each iteration, interference arising from I/Q impairments is subtracted from the received samples before updating the frequency estimate. The performance of the new scheme is assessed in terms of estimation accuracy and processing load by considering a wireless local area network (WLAN) compliant with the IEEE 802.11a standard. The main goal is to check whether the SAGE approach exhibits some advantages compared to existing alternatives.