An optimization algorithm based on improved salp swarm is proposed to optimize the orthogonal phase-coded waveform of the multiple-input multiple-output (MIMO) radar. The Logistic-Tent chaotic map is introduced into this algorithm to initialize the population, thus increase the probability of obtaining the optimal initial solution position. Then, the leader update equation is improved to expand the range of the optimal solution search. The sine function is applied for weight design, and the followers’ updated positions are flexibly adjusted according to the cost function value to further refine the convergence accuracy. The experimental results show that the auto-correlation sidelobe peak and cross-correlation peak of the phase-coded waveform of the MIMO radar designed based on the improved salp swarm algorithm are reduced by 2.6231 dB and 1.2572 dB, respectively, compared with those of the original algorithm. The orthogonal performance and the convergence accuracy of the algorithm reported here are better than those of the existing results.