Systematic investigations of the cognitive challenges in completing time diaries and measures of quality for such interviews have been lacking. To fill this gap, we analyze respondent and interviewer behaviors and interviewer-provided observations about diary quality for a computer-assisted telephone-administered time diary supplement to the U.S. Panel Study of Income Dynamics. We find that 93%-96% of sequences result in a codable answer and interviewers rarely assist respondents with comprehension. Questions about what the respondent did next and for how long appear more challenging than follow-up descriptors. Long sequences do not necessarily signal comprehension problems, but often involve interviewer utterances designed to promote conversational flow. A 6-item diary quality scale appropriately reflects respondents' difficulties and interviewers' assistance with comprehension, but is not correlated with conversational flow. Discussion focuses on practical recommendations for time diary studies and future research.