Objectives Nationwide, hospitals struggle to maintain specialist on-call coverage for emergencies. We seek to further understand the issue by examining reliability of scheduled coverage and the role of ad hoc coverage when none is scheduled. Methods An anonymous electronic survey of all emergency department (ED) directors of a large state. Overall and for 10 specialties, respondents were asked to estimate on-call coverage extent and "reliability" (frequency of emergency response in a clinically useful time frame: 2 hours), and use and effect of ad hoc emergency coverage to fill gaps. Descriptive statistics were performed using Fisher exact and Wilcoxon sign rank tests for significance. Results Contact information was obtained for 125 of 167 ED directors. Sixty responded (48%), representing 36% of EDs. Forty-six percent reported full on-call coverage scheduled for all specialties. Forty-six percent reported consistent reliability. Coverage and reliability were strongly related (P <.01; 33% reported both), and larger ED volume correlated with both (P <.01). Ninety percent of hospitals that had gaps in either employed ad hoc coverage, significantly improving coverage for 8 of 10 specialties. For all but 1 specialty, more than 20% of hospitals reported that specialists are »Never», »Rarely» or »Sometimes» reliable (more than 50% for cardiovascular surgery, hand surgery and ophthalmology). Conclusions Significant holes in scheduled on-call specialist coverage are compounded by frequent unreliability of on-call specialists, but partially ameliorated by ad hoc specialist coverage. Regionalization may help because a 2-tiered system may exist: larger hospitals have more complete, reliable coverage. Better understanding of specialists' willingness to treat emergencies ad hoc without taking formal call will suggest additional remedies.