This paper proposes a Quality of Experience (QoE) test methodology for adaptive streaming service. We analyze the factors influencing the QoE of adaptive streaming, and evaluate the QoE of the end-to-end service. Our research gives the evidence that adaptive streaming improves end-users' subjective perception greatly compared with fixed-rate streaming in terms of QoE. Two groups of experiments are conducted, one assesses artificially spliced video samples in variable bitrates, and another evaluates Dynamic Adaptive Streaming over HTTP (DASH) [1] [2] video samples under real network traces. The former experiment focuses on the influence from the bitrate distribution without considering the fluency. And the latter pays more attention to the influence of the fluency on QoE performance. Through a large number of subjective assessments, the QoE results are obtained to illustrate various impact factors on adaptive streaming, such as fluency, bitrate distribution, startup bitrate level, bitrate switching frequency, etc. The results of this study can be used to facilitate the research on mathematical modeling of user subjective experience and the algorithm development for adaptive streaming.