There has been a general trend in medicine towards greater sophistication in research design. In order to assess this trend in emergency medicine we compared the characteristics of abstracts presented at the 1974, 1983, 1989, and 1997 annual scientific meetings of academic emergency medicine. All 870 abstracts were reviewed by 1 of 3 investigators who determined research design attributes using a standardized classification scheme that has good interrater reliability. Over the last 25 years the following trends were noted: more surveys (0% v 1% v 3% v 8%, P = 0.002), more randomized studies (0% v 10% v 12% v 15%, P = 0.05), and more blinded studies (0% v 7% v 5% v 11%, P = 0.01). Tests of statistical significance were reported with increasing frequency (8% v 26% v 59% v 69%, P < 0.001) as were power calculations (0% v 0% v 1% v 3%, P = 0.02). During the study period there were also increases in the median number of authors, proportion of foreign lead authors, and the proportion of studies involving human subjects. These results reflect considerable improvement in the degree of research design sophistication reported in selected abstracts of academic emergency medicine over the study period. Further strategies to assure continued enhancement of emergency medicine research should be explored.