Multilevel researchers often gather individual-level data to measure group-level constructs. Within-group agreement is a key consideration in the measurement of such constructs, yet antecedents of within-group agreement have been little studied. The authors found that group member social interaction and work interdependence were significantly positively related to within-group agreement regarding perceptions of the work environment. Demographic heterogeneity was not significantly related to within-group agreement. Survey wording showed a complex relationship to agreement. Both evaluative items and socially undesirable items generated high within-group agreement. The use of a group rather than individual referent increased within-group agreement in response to descriptive items but decreased within-group agreement in response to evaluative items. Items with a group referent showed greater between-group variability than items with an individual referent.