We assessed the feasibility of clonality analysis with human androgen receptor gene polymerase chain reaction in terms of the sensitivity and specificity for normal and cancerous colonic tissues taken from fourteen informative cases selected from 22 women with colonic adenocarcinoma. Ten crypts microdissected from each 10-microm-thick cryostat sections and whole tissues were used as samples. DNA was extracted from the samples and amplified with and without prior enzyme digestion. These products were analyzed by capillary electrophoresis for clonality. Of the whole-tissue DNA, none of the normal tissues and seven (50.0%) of the cancerous tissues showed monoclonality. Of the microdissected samples, monoclonality was found in 88.4% (107/121) of normal crypts and 95.9% (117/122) of cancerous crypts. Samples composed of crypts with short and long alleles were found in eight of the 14 normal colonic mucosae, but in none of the cancerous tissues. We concluded that the sensitivity of this method is limited for both whole-tissue DNA and microdissected-tissue DNA, because monoclonality from small samples does not always indicate monoclonality of the entire lesion. The high specificity of this method, however, allows polyclonal results in whole tissues to be confirmed by additional analysis of microdissected tissues.