For reliable epidemiological studies and effective control programs, accurate detection of Trypanosoma cruzi in vectors is essential. Previously, T. cruzi has been detected by microscopic examination, however, more recently, PCR analysis of vector fecal samples has been shown to be an effective means of detection in particular vectors. We compared the sensitivities of PCR and microscopy for detection of T. cruzi in different anatomical locations of the two major vectors of Guatemala, Triatoma dimidiata and Rhodnius prolixus. One hundred thirty five bugs were collected from houses in six rural towns in five departments and dissected. Samples were obtained from the rectal gland, intestines, stomach and salivary glands and analyzed by microscopy and PCR. T. cruzi was detected in T. dimidiata and R. prolixus. T. rangeli, a nonpathogenic but morphologically similar hemoflagellate, was found only in R. prolixus. A sample was considered positive by microscopy if any hemoflagellate was observed (T. cruzi and T. rangeli). Preliminary studies established that the assay is biased for detection of T. cruzi. T. cruzi can be detected in the presence of 90% T. rangeli. However, T. rangeli must make up at least 75% of the sample before it is detected. For T. dimidiata rectal samples, PCR was somewhat more sensitive than microscopy (39.1% and 24.6% positive, respectively). This difference was not significant.  In R. prolixus, PCR proved more than twice as sensitive as microscopy: 57.6% positive by PCR as compared to 22.7% by microscopy for rectal samples.  Rectal samples showed the highest rates of infection followed by intestine, and stomach samples.   In R. prolixus, analysis of the rectal sample alone was not sufficient as 10.5% of the Rhodnius infections would have been missed if only the rectal sample had been analyzed. More than 50% of the bugs collected in the department of Zacapa were infected, the highest rate of infection of the four departments sampled: Guatemala, Santa Rosa, Chiquimula, and Zacapa. Thus, PCR is significantly more sensitive than microscopy for detection of T. cruzi for R. prolixus but not T. dimidiata. Analysis of anatomical locations in addition to the rectal sample may be necessary for accurate assessment of infection in particular vectors.