The main aim of this study was to investigate which factors are likely to affect the probability that events of violent collective action are reported by the press in KwaZulu-Natal. The study hypothesised that the likelihood of violent conflict events being reported by the press is affected by certain characteristics of the events themselves, such as their intensity and size, and by the environment in which events occur, such as their physicai location and the prevailing political context. The study was limited to the KwaZulu-Natal province where levels of violent collective action have been the highest in the country over the past decade. This province is also home to many violence monitoring agencies, which constituted an important alternative source of information against which the reporting trends of the newspapers in the province could be compared. The main source of information used in this study was the Conflict Trends in KwaZulu-Natal project's database of collective action events, which comprises events reported by both the press and the monitoring agencies. Data on a total of 3990 violent conflict events was analysed during 1987, 1990 and 1994, in the form of comparisons between the reporting tendencies of the press and the monitors. Interviews were also conducted with reporters and editors of the daily newspapers in KwaZulu-Natal, as well as with selected monitoring agency staff members. These interviews provided valuable information about how these media operate, and the factors which constrain their violence reporting activities. The most support was found for the argument that the political context influences violence reporting by the press. Analyses suggested that in all three years studied, the press contribution to the database decreased as monthly levels of violent conflict increased. This was explained in relation to the prevailing political context. The results also showed that reporting trends changed over time. In the earlier years, the press did not appear to be more inclined to report events of larger size and intensity, or events which were close to the newspapers' base. In 1994, however, this trend was reversed. There also was no clear evidence that the States of Emergency impacted negatively on press reporting of violence in terms of the variables studied. In addition, the study concluded that both the press and the monitOring agencies had made important separate contributions to the database on violent collective action. It is, therefore, vital that systematic studies of violent conflict in KwaZulu-Natal make use of multiple sources of data. / Thesis (M.A.)-University of Natal, Durban, 1995.