Summary.? Survey agencies have started to use interviewer observations collected on all sample units to adjust survey estimates for non-response. Ideally, these observations should be related to both response indicators and key survey variables. However, these observations are typically judgements that are made by the interviewers, making them potentially prone to measurement error. Presenting analyses of data from the National Survey of Family Growth in the USA, this study examines the quality and utility of these interviewer observations and considers the implications of measurement errors in these observations for the effectiveness of non-response adjustments.