OBJECTIVES: Secondary data sources can provide a wide variety of useful data for research studies, but their utility depends on their validity.  It is unclear whether the extent to which secondary sources are used differs between research-oriented and practice-oriented journals.  The objectives of this study were to describe the proportion of original research articles using secondary data, the types of secondary data being used, and whether validation was mentioned while comparing these findings between two types of journals. METHODS: All original research articles published in 2010 from selected practice-oriented journals (New England Journal of Medicine, Archives of Internal Medicine, Pharmacotherapy, Annals of Pharmacotherapy) and research-oriented journals (Value in Health, Medical Care, Health Services Research) were reviewed.  The complete text of each article was examined to determine whether secondary data were used, the type of secondary data used, and whether validation of the data was mentioned.  The proportion of original research articles using secondary data and the mention of validation was compared between the two types of journals using Fisher’s exact test. RESULTS: There were 192 and 425 original research articles for the research- and practice-oriented articles, respectively.  Overall, 44.1% of research articles used secondary data.  Articles in research-oriented journals were more likely than practice-oriented journals to use secondary data (69.8% vs. 32.5%, respectively; p <0.001) and to mention validation (17.2% versus 5.8%, respectively; p = 0.003).  Administrative data were the most common secondary data source in the research journals compared to medical records in the practice journals. CONCLUSIONS: The use of secondary sources was more common in research-oriented journals than in practice-oriented journals.  A variety of secondary sources were used in published articles.  Validation was mentioned more frequently in research journals than in practice journals, but the overall percentage was relatively low for both types of journals.