Basic methodological issues involved in field research are described, and criteria are presented for conducting field studies from which valid generalizations can be made regarding the accuracy of polygraph tests in criminal investigations and judicial applications. The existing field studies are evaluated in light of those criteria and are found to be insufficient to provide firm estimates of polygraph test accuracy in field contexts. Even if the overall accuracy of such tests is high, when baserates of deception are relatively low or high there are additional problems regarding confidence in the obtained outcomes. A methodology is presented to estimate baserates from test outcomes obtained in applied settings and to calculate confidence in truthful and deceptive outcomes. The method is illustrated with examples from law enforcement, judicial proceedings, and personnel screening. It is suggested that polygraph test accuracy in field settings should be estimated with caution, bearing in mind the limited data available from methodologically adequate field studies and the problems which arise from extreme baserates in some applied contexts. Three tables, equations, 31 references. (Author abstract modified).