Background Self‐reported consumption is pervasive in alcohol research, though retrospective recall bias is a concern. Fine‐grained methods are designed to limit retrospection; yet, discrepancies can arise when comparing responses on fine‐grained surveys with responses to retrospective surveys across weeks or months. Many fine‐grained studies use both repeated daily surveys (RDS) and end‐of‐day (EOD) summaries, but little research has examined whether these survey types are consistent. The purpose of this study was to quantify the magnitude and directionality of discrepancy between EOD summaries and RDS and identify alcohol‐related predictors of discrepancy. Methods As a part of a larger study, college student alcohol and cannabis users (N = 341; 53% women; Mage = 19.79 years) were recruited to complete 56 days of data collection, including 5 daily assessments of their substance use and related constructs, one of which included an EOD summary of the previous day. Generalized linear mixed‐effects models were used to examine between‐ and within‐person predictors of a 5‐category, discrepancy outcome: no discrepancy, low discrepancy where RDS < EOD, low discrepancy where EOD < RDS, high discrepancy where RDS < EOD, and high discrepancy where EOD < RDS. Results Discrepancies between EOD and RDS were observed in both directions. Alcohol problems predicted more alcohol consumption reported on the EOD survey than across RDS. Within‐person alcohol quantity and hourly rate of consumption were most strongly related to less alcohol consumption reported on the EOD survey. Between‐ and within‐person peak subjective intoxication and within‐person liquor consumption were associated with discrepancies in both directions. Conclusions Surveys requiring more retrospection may overestimate alcohol consumption in problematic drinkers and underestimate consumption on days where more alcohol is consumed than typical. Evidence also suggests that greater day‐to‐day instability in alcohol behavior is linked to less consistent reporting overall. More research is needed to discern factors contributing to inconsistent reporting on fine‐grained surveys to maximize the validity of reports.