This paper presents an adaptive multiple targets detection framework for frequency diverse array multiple-input multiple-output radar embedded in Gaussian noise with an unknown covariance matrix. To this end, we define the one-range-cell multiple targets model as a summation expression and then design four detectors, i.e., the generalized likelihood ratio test, adaptive matched filter, Rao, and Wald tests for the above newly built problem. Closed-form expressions for the probability of false alarm and the probability of detection are provided to assess the aforementioned detectors. The numerical examples show that the proposed architectures can ensure better detection performance than the considered competitors. Finally, notice that each proposed detector has a specific behavior in terms of robustness to low volumes of data. Thus, the choice of a specific solution is dictated by the operating requirements of the radar system.