The study of conflict prevention has only recently become moresystematic and moved away from an exclusive reliance on casestudies. With the efforts of the Uppsala Conflict Data Program(UCDP) and the Folke Bernadotte Academy, a dataset of thirdparty measures taken in armed conflicts at low intensity hasbeen created. The dataset, ‘Managing Intrastate Low-level Conflicts’(MILC) covers all existing low-level armed conflicts for theperiod of 1993-2004.There is a lingering discussion regarding whether systematicallycollected data based on news media sources, other currentreports and general insight actually captures the events thatdiplomats, mediators and other actors would regard as significant.Some would even question if systematic data like the MILCdataset can generate an accurate and unbiased account of thephenomena it tries to quantify. This study is responding to theseconcerns by comparing news based information with informationbased on memoirs and historical accounts.Analyzing the case of Burundi using both approaches, the purposeof this study is to see if different approaches will result ina similar or different understanding of third party activities ina particular crisis. Comparing the MILC dataset with data collectedin other ways and using other types of material will helpus understand the strengths and limitations of a systematic approachbased on news reports. This in turn will provide a betterfoundation for interpreting and analysing this systematic data.