In this paper, we compare the performance of multiple-input-multiple-output (MIMO) techniques applied to indoor optical wireless communications (OWC) assuming line-ofsight (LOS) channel conditions. Specifically, several 4 × 4 setups with different transmitter spacings and different positions of the receiver array are considered. The following MIMO algorithms are considered: Repetition Coding (RC), Spatial Multiplexing (SMP) and Spatial Modulation (SM). Particularly, we develop a framework to analytically approximate the bit error ratios (BERs) of these schemes and verify the theoretical bounds by simulations. The results show that due to diversity gains, RC is robust to various transmitter-receiver alignments. However, as RC does not provide spatial multiplexing gains, it requires large signal constellation sizes to enable high spectral efficiencies. In contrast, SMP enables high data rates by exploiting multiplexing gains. In order to provide these gains, SMP needs sufficiently low channel correlation. SM is a combined MIMO and digital modulation technique. We show that SM is more robust to high channel correlation compared to SMP, while enabling larger spectral efficiency compared to RC. Moreover, we investigate the effect of induced power imbalance between the multiple transmitters. It is found that power imbalance can substantially improve the performance of both SMP and SM as it reduces channel correlation. In this context, we also show that blocking some of the links is an acceptable method to reduce channel correlation. Even though the blocking diminishes the received energy, it outweighs this degradation by providing improved channel conditions for SMP and SM. For example, blocking 4 of the 16 links of the 4 × 4 setup improves the BER performance of SMP by more than 20 dB, while the effective signal to noise ratio (SNR) is reduced by about 2 dB due to the blocking. Therefore, MIMO techniques can provide gains even under LOS conditions which - rovide only little channel differences.