Adopting extremely low-resolution (e.g. one-bit) analog-to-digital converters (ADCs) and digital-to-analog converters (DACs) is able to bring a remarkable saving of low-cost and circuit power for multiple-input multiple-output (MIMO) radar.In this paper, the problem of joint design of transmit waveform and receive filter for collocated MIMO radar with a architecture of one-bit ADCs and DACs is investigated. Under this architecture, we derive the output quantized signal-to-interference-plus-noise ratio (QSINR), which is relative to the detection performance of target, in the presence of signal-dependent interference. The optimization problem is formulated by maximizing the QSINR with a binary waveform constraint. Due to the nonconvex objective and binary constraint, the resulting problem is hard to be directly solved. To this end, we propose an alternating minimization algorithm. More concretely, at each iteration, the closed-form solution of the receive filter is attained by exploiting the minimum variance distortionless response (MVDR) method, and then the one-bit waveform is optimized with the aid of the alternating direction method of multipliers (ADMM) algorithm. In addition, the performance gap between the one-bit MIMO radar and infinite-bit MIMO radar is theoretically analyzed under the noise-only case. Several numerical simulations are provided to demonstrate the effectiveness of the proposed methods.