From cross-layer perspective, the impact of imperfect spectrum sensing and access contention on the cooperative sensing in cognitive radio networks is investigated in the context of the tradeoff between interferences to PUs and aggregated secondary throughput. Jointly considering imperfect spectrum sensing and access contention, we reformulate the sensing-throughput tradeoff problem via taking the interference probability, rather than the detection probability, as the optimization constraint, and further obtain the optimal combination of fusion rule, sensing duration and detection threshold to maximize the secondary throughput under the interference probability constraint. Numerical results show that the proposed cross-layer method can improve the secondary throughput performance significantly, especially in the case of low signal-to-noise ratio.