In a series of 2,665 frozen sections, false-positive diagnoses of cancer occurred only four times (0.15 percent) and false-negative diagnoses of cancer 43 times (1.61 percent). In 46 instances, the diagnosis had to wait for paraffin sections.

Errors of three types: sampling, interpretive, and communicative — can occur, and sampling errors account for more than half of the errors. The types of lesions which lead to interpretive errors are the very well differentiated malignant tumors which are confused with benign proliferative conditions and the very poorly differentiated malignant tumors which are confused with inflammation. No significant errors of communication occurred in this series.

The only indication for a frozen section is the need for an immediate decision requiring immediate action. These decisions may revolve about the nature of the disease, the extent of the disease, biopsy technique, identification of tissue, viability of tissue, or even administrative matters. If a decision will be made or a course of action selected, regardless of the frozen section results, one should not get a frozen section.