The progesterone receptor (PR) is an important marker of response to endocrine agents in breast cancer. Immunohistochemical demonstration of PR in formalin fixed tissue has previously proved difficult, and heat pretreatment is considered necessary to retrieve the antigen. There are few data on the effectiveness of autoclaving in unmasking PR, however, and it is not known whether all PR epitopes are equally unmasked. The objectives of this study were to compare the efficacy of autoclaving and microwaving to retrieve PR antigen in archival breast tumors, to determine whether there is an epitope-dependent variability in the pretreatment required, and to examine different slide types and adhesives to reduce the problem of section loss frequently associated with these procedures. Paraffin embedded sections were cut at 2 or 4 microm, mounted onto various slide types with or without the addition of adhesive, and heat pretreated prior to immunoperoxidase staining. Whereas PR immunoreactivity was clearly demonstrated in tissue after both autoclaving and microwaving, autoclaving produced a significantly stronger staining intensity under the conditions used in this study. The duration of autoclaving required to reveal PR fully differed for different epitopes examined. In the absence of heat pretreatment, PR was not detected. Section retention was improved by the use of adhesives and by cutting tissue at 2 microm. Maximum retention was obtained using positively charged slides coated with Mayer albumen adhesive. We conclude that for maximal tissue preservation autoclave pretreatment is the preferred method of PR antigen retrieval from archival breast tumors, that there is epitope-dependent variability in pretreatment required, and that section loss during this procedure can be minimized by choice of slide type, the use of adhesive, and by cutting sections at 2 microm.