— Semi-blind spatial equalisation is considered for multiple-input multiple-output (MIMO) systems that employ high-throughput quadrature amplitude modulation scheme. A minimum number of training symbols, equal to the number of transmitters, are first utilised to provide a rough least squares channel estimate of the system’s MIMO channel matrix for the initialisation of the spatial equalisers’ weight vectors. A constant modulus algorithm aided soft decision-directed blind algorithm is then employed to adapt the spatial equalisers. This semiblind scheme has a very-low computational complexity, and it converges fast to the minimum mean-square-error spatial equalisation solution as demonstrated in our simulation study.