Integrated-test tasks, which combine receptive and productive language skills in task performance, e.g., listening-speaking or listening-reading-speaking, are increasingly being used in second language assessment, including in high-stakes English exams such as the TOEFL iBT and PTE Academic. Although recent studies (Plakans, 2008; Sawaki, Quinlan, & Lee, 2013) have found that the construct of each individual skill involved in task performance (e.g., listening, reading, and writing) is present and distinct, it is not entirely clear what abilities are actually assessed by the tasks, especially as far as listening is concerned. This study thus analysed test-takers’ listening comprehension processing behaviours while completing listening-to-summarise tasks. In addition, test-takers’ perceptions of the tasks and of listening task difficulty were investigated. The aim of this was to be able to describe the listening construct measured by integrated-listening tasks.Data was collected from 72 Thai English as a Second Language (ESL) learners. Each participant completed four listening-to-summarize tasks – two tasks requiring an oral summary and two a written summary. To investigate the comprehension processing behaviours performed to complete the tasks, a stimulated recall was conducted with 12 participants after each task. To study the perceptions of the tasks and of listening difficulty and their relation to task performance, the remaining 60 participants completed a perception questionnaire after each task. The results showed that to comprehend listening input with the aim of summarizing it, the participants engaged in both lower-level and higher-level cognitive processes and these cognitive processes were facilitated and monitored by a number of strategies. However, to maintain focus on the text’s main point and accurately understand it, it was necessary that the participants successfully activated comprehension monitoring, real-time assessment of input, and lower-lever cognitive processes. Lack of the successful application of these processes and strategies often led to misinterpretations of the text, partly because of the interference of background knowledge which was not congruent with the texts’ information. Participants with different performance levels were found to engage in different types of processes and strategies, with different degrees of success. The participants, in addition, were found to perceive the tasks as authentic and a fair way to assess their abilities to use English for academic purpose, especially listening abilities. In addition to providing a description of the listening construct measured by integrated-listening tasks, the study suggests that listening comprehension ability should be integrated in the description of the task construct and both cognitive and strategic processing should be recognized as part of the construct. On the basis of the findings, a model of second language (L2) listening in the context of listening-to-summarize tasks is formulated.