This study deals with the problem of joint design of multiple-input–multiple-output (MIMO) radar transmit waveform and receive filter in the presence of signal-dependent interferences. To enhance the detection performance of a slow moving target, MIMO radar space-time adaptive technology is adapted. Considering the issue of imprecise knowledge about target's Doppler frequency and direction, the worst-case output signal-to-interference-plus-noise ratio over the uncertainty set of target steering vector is set as figure of merit. Furthermore, due to the non-linear power amplifier and the fact that transmit antenna may be out of work, constraints of peak-to-average power ratio and transmit power for each transmit antenna are taken into account. As to the resulted hard problem, some suitable mathematical transformations are performed and a novel sequential optimisation procedure combined with semi-definite relaxation, Charnes–Cooper transform and Lagrange duality method is developed to tackle the non-convex joint design problem. Randomisation procedure is utilised to synthesise the final transmit waveform and receive filter. The computational complexity is linear with the numbers of iterations and trials of the randomisation. The convergence of the devised algorithm is analysed and numerical results in a variety of scenarios are provided to demonstrate the effectiveness of the proposed algorithm.