Objective: The objective of this study was to determine why different criteria for response to treatment of bacterial vaginosis (BV) led to markedly different conclusions about treatment efficacy in a randomized trial comparing metronidazole gel versus metronidazole/nystatin ovules. Study: We compared the impact of two treatment regimens on individual components of Amsel and Nugent criteria at follow-up visits 14, 42, and 104 days after initiating treatment. Results: Compared with gel, ovules more effectively eliminated amines, clue cells, and Gardnerella, Prevotella, or Mobiluncus morphotypes from vaginal fluid, thus achieving cure based on "usual" criteria (absence of BV by Amsel or Nugent criteria), but did not more effectively restore Lactobacillus morphotypes or lower vaginal pH, thus not meeting Federal Drug Administration (FDA) criteria for cure. Conclusion: Because early vaginal recolonization by lactobacilli was poor after both gel and ovules, FDA draft criteria for cure missed marked differences in treatment efficacies against Gardnerella, clue cells, and amines. Cure defined more "usually" may give more useful information. Copyright © 2005, American Sexually Transmitted Diseases Association All rights reserved.