This paper proposes a new fuzzy optimization approach for the quality of service (QoS)-aware web service selection (QWSS) problem under uncertain QoS parameters. To deal with the ambiguity phenomena of some web services’ QoS characteristics like the execution time attribute, which strongly fluctuates depending on the rates of network traffic and systems congestion; hence, the generalized trapezoidal fuzzy number (GTrFN) is used to represent the uncertainty of the QoS values. Subsequently, the QWSS problem with GTrFN-based QoS proprieties is modeled as a fuzzy optimization (FQWSS) problem. To solve the FQWSS model, we design a new fuzzy teaching learning based optimization (FTLBO) algorithm, for which a local search method and an elitism operator are embedded into it to enhance the exploitation ability and speed up the convergence of FTLBO. The experiment results of comparing the FTLBO algorithm to some recent fuzzy optimization approaches demonstrate its effectiveness and performance in solving the FQWSS problem with different scales.