This study focuses on investigating the computerized medical record, including textual progress notes, using data and text mining techniques to examine patient fall-related injuries (FRIs) in the Veterans Administration (VA) ambulatory care setting. FRIs are high cost, high volume adverse events in the VA that are difficult to identify from VA administrative databases. Recognizing patterns in progress notes can aid in identifying those records that should have been coded as FRIs in the administrative data. This facilitates understanding the frequency and nature of fall related injuries for implementation of prevention programs at the VA. Latent semantic indexing is used to create structured data from large textual fields found in the medical records. Unsupervised learning (cluster analysis) is used to assess the potential predictive power of the textual descriptions. Two predictive data mining approaches are then used, in combination with supervised text mining, to classify patient records as fall-related injuries.