This study was prompted largely by criticisms that magistrates' courts have a high conviction rate in comparison to jury trials in crown courts, because magistrates have a bias favoring the police and the prosecution. This study examined the evidentiary content of cases contested before magistrates in a sample of observed trials at which defendants plead not guilty. The trials occurred in selected magistrates' courts from January to June 1979. Six courts which try a large number of cases were selected from different regions of the country. A total of 394 charges were contested by 318 defendants. Observers' checklists elicited information about the main elements of the prosecution and defense cases. Predictably, direct evidence from prosecution witnesses whose credibility was not impugned was associated with a high level of conviction. Conversely, the analysis also confirmed a strong association between acquittal and factors pointing to weaknesses in the prosecution case, notably lack of prosecution witness credibility and the absence of confessions or direct evidence implicating the defendant. A significant finding, however, was that where defendants' credibility was not manifestly impugned, 63 percent of the charges still resulted in conviction. In most of these cases, there was no confession, but there was direct evidence (primarily police eyewitness evidence) describing the defendant's conduct and the criminal intent implicit in the observed behavior. There was an apparent readiness in the magistrates to accept the police version of disputed events against the unimpugned testimony of defendants. This finding calls into question the objectivity of magistrates in considering the realiability of eyewitness testimony, which research has shown to be prone to error, whether or not the eyewitness is a trained observer. Appended are discussions of the assessment of witness credibility and log-linear modeling. Tabular data and 20 references are provided.