Testing under the actual operational profile (OP) for high-reliability software can be expensive, time consuming or even infeasible in situations where performance of a system is dominated by infrequent but highly critical events. To solve this problem, testing using importance sampling was put forward for usage-based software reliability demonstration testing (SRDT). An extended research based on OP for software reliability growth testing (SRGT) is proposed. Firstly, the reinforced OP based on changing the probabilities of critical operations is suggested. Secondly, the principle of accelerated SRGT is proposed. Then an implementing framework is proposed. Finally, the accelerated method is applied on a software system. The experimental results indicate: compared with the growth testing approach based on original OP, the accelerated growth testing approach based on the reinforced OP can not only significantly decrease the required testing time and the required test cases, but also obtain the same testing conclusion and the unbiased estimation result.