The agricultural industry has been ranked among the most hazardous. Yet, it has been alleged that occupational injuries and fatalities are seriously underreported. Access to quality agricultural injury data poses a special problem. The purpose of this study was to investigate the concordance of reporting of fatal agricultural injuries between death certificate data and the Newspaper Clipping Service data in the state of Minnesota, utilizing a surveillance system developed by the University of Minnesota. Between September 1, 1981, and August 31, 1986, a total of 350 agricultural fatalities were identified in Minnesota; 82% were identified through death certificate data and 67% through the Newspaper Clipping Service. Differences in reporting between the two data sources were noted for gender, age, injury type, anatomical site, source, mechanism of injury, and multiple versus single injury. If only death certificates had been utilized, 18% of the fatalities would have been missed. Although it is apparent that death certificate data have an advantage over the Newspaper Clipping Service data for fatality reporting of specific variables, this study revealed that death certificate surveillance alone will miss mortality data and detection of certain potential risk factors. Suggestions for improving surveillance of agricultural fatalities are identified. However, until relevant changes are made, it will be essential to use a combination of data sources that include the Newspaper Clipping Service to identify agricultural injury fatalities accurately.