Even though the interconnected subsystems of a class of nonlinear systems are controlled by the same sampling period, it is common that the sampled-data time instant of each subsystem is different. In this case, the problem of using sampled-data output feedback to design global decentralized stabilizers for the decentralized systems is quite challenging but also interesting. To solve this problem, in this paper we propose a methodology to construct a set of decentralized stabilizers, which are implemented using the unsynchronized sampling instants.