This paper presents a practical virtual multiple-input-multiple-output (MIMO) system that implements bit-interleaved coded modulation (BICM) and compress-and-forward (CF) cooperation. A minimum mean square error (MMSE) receiver is considered since it has low complexity and allows good performance when combined with BICM techniques. A closed-form upper bound for the system error probability is derived, and based on this we prove that the smallest singular value of the cooperative channel matrix determines the system error performance. Accordingly, an adaptive modulation and cooperation scheme is proposed, which uses the smallest singular value as the threshold strategy. Depending on the instantaneous channel conditions, the system could therefore adapt to choose a suitable modulation type for transmission and an appropriate quantization rate to perform CF cooperation. It is shown that the adaptive modulation and cooperation scheme not only enables the system to achieve comparable performance to the case with fixed quantization rates but eliminates unnecessary complexity for quantization operations and conference link communication as well.