Pixelated optical wireless systems can transmit at high speed by encoding data in transmitted images using spatial orthogonal frequency division multiplexing (spatial OFDM) but are susceptible to a number of limiting factors including defocus, linear fractional misalignment (LFM) and vignetting. This paper investigates the effects of these impairments on the performance of both forms of spatial OFDM: spatial DC-biased optical OFDM (SDCO-OFDM) and spatial asymmetrically clipped optical OFDM (SACO-OFDM). The novelty of this paper is in the theoretical analysis of the effect of defocus on a spatial OFDM based pixelated system. The theoretical results for defocus are found to be in agreement with the simulation results reported in the literature. Furthermore, the joint effects of defocus, LFM and vignetting are mathematically described. Analytical and simulation results show that these three factors are largely independent of each other. Simulation results indicate that these impairments and their associated mitigation techniques influence the optimum choice of DC bias for SDCO-OFDM. Finally, it is shown that for the case of the joint impairments and for a given data rate, the bit error rate performance of SDCO-OFDM with the optimum bias is slightly better than that of SACO-OFDM. ], which is an extension of the concept of conventional OFDM [18][19][20] to the 2-D spatial domain. For spatial OFDM, data are encoded in the spatial-frequency domain, the frequency representation of the 2-D image space, using a number of orthogonal 2-D subcarriers. Each of the encoded spatial frames is converted to a pixelated image before transmission. Spatial OFDM has several advantages for pixelated communication. Compared with systems that encode data directly in the spatial domain, a spatial OFDM scheme is less affected by a number of distortions occurring in the spatial domain [7,8,21,22]. Moreover, the formation of the data-carrying transmitted images can be well-designed. This means that, by using spatial OFDM algorithms at the transmitter side, the images can be generated in a way that makes them much robust to distortions. The receiver can then decode the