Institutional review boards (IRBs) are almost universally considered over-worked and under-staffed, at the same time requiring substantial commitments of time and resources. Although some surveys report average IRB memberships of 15 persons or more, federal regulations require only five. We present data on IRB meetings at 8 of the top 25 NIH-funded academic medical centers in the U.S., indicating substantial contributions from primary reviewers and chairs during protocol discussions but little from other members. The implications of these data for current IRB functioning are discussed and an alternative model proposed.