It is a well-known fact that tubal stenosis and/or peritubal adhesion are often associated with Chlamydia trachomatis infection. Although tubal pregnancy may be attributed to this infection, there are only extremely rare cases in which the presence of C. trachomatis has been confirmed by immumo-histochemical examination on tissues isolated from patients with tubal pregnancy. We thus tried to confirm the presence of C. trachomatis infection by detecting DNA of the organism in tissues surgically isolated from patients with tubal pregnancy. Two detection methods, a ligase chain reaction (LCR) method and an immuno-histochemical staining which detects an antigen of C. trachomatis, were compared using paraffin-embedded tissue samples which were surgically isolated from patients with tubal pregnancy or hydrosalpinx. The LCR method was capable of detecting DNA of C. trachomatis in tissue samples in which the C. trachomatis-specific antigen could not be detected using immuno-histochemical staining. This was due to the fact that immuno-histochemical staining methods are applicable to the analysis of sequences the length of which range from 200 to 400 bp (base pairs), while the LCR method used here allows the analysis of sequences as small as 48 bp. This fact makes the LCR method a very convenient tool, as compared with immuno-histochemical methods, for analysis of the paraffin embedded tissue samples where by effects of formalin--used for fixation for pathologic diagnosis--the risk of fragmenting the DNA samples is relatively high. Presence of C. trachomatis DNA as detected by LCR method in surgically isolated samples from patients with tubal pregnancy supports the involvement of Chlamydia trachomatis infection in the occurrence of tubal pregnancy. Accordingly the LCR method is capable of detecting DNA of C. trachomatis in paraffin-embedded samples of tubal tissue in which presence of C. trachomatis could not be confirmed by an immuno-histochemical staining method.