In a cognitive radio system, it is still a problem that resource scheduling among secondary users (SUs) is distributed unfairly when spectrum aggregation technology is involved to support high-speed data transmission. A global proportional fair scheduling algorithm is proposed based on spectrum aggregation to solve the problem. This paper focuses on the relation between the fairness for SUs and spectrum aggregation. Throughput fairness can be guaranteed as much as possible after considering two parameters, the span of spectrum aggregation and the remaining data queue length of SUs. Simulation results show that the proposed scheduling algorithm takes advantage of fairness and the delay of system service as compared with the other two scheduling algorithms, max C/I and partial proportional fairness. Meanwhile, it is shown that the proposed scheduling algorithm has lower throughput shake to make sure of fairness among SUs and high efficiency of system.