political violence if only carried out by non-state actors, but the matter becomes altogether more serious if state agents such as the police are involved. Methodology The Human Rights Forum has maintained two distinct data bases on human rights violations since 2000. The first, which has been in operation since 1999, has recorded cases of human rights violations reported in the public press; the second, which has been in operation since 2000, contains the records of actual reports of violations made to the Human Rights Forum itself of the various members of the Human Rights Forum. These two data bases were searched for cases involving allegations of violations committed by the ZRP, and then the data sets were analysed. The data was first described generally as frequencies. The data was compared in various ways according to a number of hypotheses that were derived from the findings of previous reports. Here only statistically significant findings are reported. Comparison of data bases A basic question always to be asked in any statistical analysis concerns the quality and reliability of the data. The Human Rights Forum had two data bases drawing from different sources, as indicated above. One, the press reports database, comprises reports of gross human rights violations reported in the media, and mainly the print media (“press reports”). This contains 444 reports in which the Zimbabwe Republic Police were mentioned as being involved in a human rights violation. The second, the case reports database (“Forum data”), consists of actual reports made by victims of alleged human rights violations to the Human Rights Forum or one of its members (“victim reports”). Clearly, if the data in these two databases is valid, then there should be strong agreement between the two data sets in most respects. Thus, the data from the press reports [n=444] was compared, where possible, with the data taken from the reports actually given to the Human Rights Forum [n=1237]. Four comparisons were made where possible with the gross frequencies: the frequency of reports per Province, the frequency of reports per year, the frequency of alleged perpetrators, and the frequency of alleged violations. As regards the frequency of reports per Province, it was evident that there was no difference in the distributions between the two data sets. In fact, the agreement was exceptionally strong, as can be seen from Figure 2. The majority of the reports in both data sets came from Harare, with a greater percentage from Harare being reported in the victim reports. There were very few Who guards the guards? Violations by the Zimbabwe Republic Police, 2000 to 2006