In 1989, the College of American Pathologists Q-Probes Quality Assurance Program studied intraoperative frozen-section consultations performed in 297 institutions with mean bed size of 316 (range, 0 to 1351 beds) in North America during 5 consecutive months. The aggregate database was composed of 933,751 surgical cases (mean, 3144 per institution); 52,464 frozen-section cases (mean, 177); and 79,647 individual frozen sections performed (mean, 268). The rate of frozen sections per all surgical case accessions was 5.6% (cases with frozen section) and 7.3% (individual frozen sections performed), with an average of 1.5 frozen sections per case. Frozen-section rate increased proportional to bed size, from less than 5% in institutions with bed size below 150 to 15% in institutions with bed size above 600. Of all frozen sections performed, 4.2% were deferred. Deferrals to paraffin sections in pathologists' opinions were 92.6% appropriate, 1.2% inappropriate, and 6.2% not stated. When frozen-section diagnoses were compared with permanent section diagnoses, there was a 98.3% diagnostic concordance, adjusted for deferred diagnoses, but including the performance of frozen sections on mammographically directed biopsy specimens with no gross abnormalities in 80% of institutions. This practice accounted for 11.8% of the discordant frozen-section diagnoses. The reasons for diagnostic discordances were gross tissue sampling (44.8%); misinterpretation (40%); sectioning (12.7%); inadequate history (5.6%); staining (1.5%); labeling (0.5%); and other (3%). Assessment of diagnostic discordance on patient outcomes by the reviewing pathologist showed that patient management was unaffected in 74%, minimally affected in 20%, and greatly affected in 2.5%.