Although good and timely information on agricultural production is critical for policy-decisions, the quality of underlying data is often low and improving data quality could have a high payoff. This paper uses data from a production diary, administered concurrently with a standard household survey in Uganda to analyze the nature and incidence of responses, the magnitude of differences in reported outcomes, and factors that systematically affect these. Despite limited central supervision, diaries elicited a strong response, complemented standard surveys in a number of respects, and were less affected by problems of respondent fatigue than expected. The diary-based estimates of output value consistently exceeded that from the recall-based production survey, in line with reported disposition. Implications for policy and practical administration of surveys are drawn out.