Many important national surveys use audio computer-assisted survey interviewing (ACASI) to collect sensitive information from survey respondents in face-to-face interviews. ACASI is used to mitigate concerns about education and literacy, and ensure privacy when respondents are providing answers to sensitive questions. However, human interviewers are still required to introduce the ACASI portions of interviews, describe how to use the technology, and potentially assist respondents when questions arise. In addition, interviewers may behave in different ways while the respondent is answering ACASI questions, and could vary in their ability to establish a private environment. This study examines the effects of selected interviewer behaviors during ACASI, ascertained from post-survey interviewer observations, on the response distributions for selected sensitive survey items in the National Survey of Family Growth (NSFG). This study also considers the amount of variability among NSFG interviewers in the frequency of these behaviors, finding it to be substantial. The results of the study show that these behaviors may introduce interviewer effects on sensitive survey responses. The findings presented here motivate future randomized experiments that could explore the effects of these interviewer behaviors further and inform interviewer training with respect to behaviors that should be avoided during ACASI data collection.