Due to its time-varying nature, oscillator phase noise can significantly degrade the performance of channel estimation, carrier recovery, and data detection blocks in high-speed wireless communication systems. In this paper, we analyze joint channel, carrier frequency offset (CFO), and phase noise estimation plus data detection in orthogonal frequency division multiplexing (OFDM) relay systems. To achieve this goal, a detailed transmission framework involving both training and data symbols is presented. In the data transmission phase, a combtype OFDM symbol consisting of both pilots and data symbols is proposed to track phase noise over an OFDM frame. Next, a novel algorithm that applies the training symbols to jointly estimate the channel responses, CFO, and phase noise based on the maximum a posteriori criterion is proposed. Additionally, a new hybrid Cramér-Rao lower bound for evaluating the performance of channel estimation and carrier recovery algorithms in OFDM relay networks is derived. Finally, an iterative receiver for joint phase noise estimation and data detection at the destination node is derived. Extensive simulations demonstrate that the application of the proposed estimation and receiver blocks significantly improves the performance of OFDM relay networks in the presence of phase noise.