OBJECTIVE: Study the reproducibility of wet smear interpretation of clue cells, lactobacillary grades and leukocyte dominance with conventional bright light and phase contrast microscopy. STUDY DESIGN: Sets of vaginal specimens were taken from unselected consecutive women attending an outpatient gynaecology clinic. Air-dried vaginal fluid on a microscope slide was rehydrated with isotonic saline before examination by six independent international investigators. Some investigators initially used a conventional bright light microscope, followed by phase contrast technique. RESULTS: Using phase contrast microscopy, an excellent inter-observer agreement was obtained among all investigators for clue cells detection (Kappa values from 0.69 to 0.94) and lactobacillary grades (Kappa 0.73-0.93). When conventional light microscopes were used, poor agreement was obtained for these criteria (Kappa index 0.37-0.72 and 0.80, respectively), but switching to phase contrast microscopy by the same investigators, improved Kappa to 0.83-0.85 and 0.88, respectively. The inter-observer agreement for estimation of the leukocyte/epithelial cell ratio (Kappa index 0.17-0.67) was poor, irrespective of the type of microscopy applied. Intra-observer agreement of clue cell detection and lactobacillary grading was also found to be excellent if phase contrast microscopy was used (Kappa 0.87-0.93), and poor with conventional bright light microscopy (Kappa 0.45-0.66). CONCLUSION: Clue cells and the lactobacillary grades are reliably identified by phase contrast microscopy in wet smears, with excellent intra- and inter-observer reproducibility agreement, and better than when simple bright light microscopy was used. Evaluation of leukocyte grading, on the other hand, was inconsistent among the different microscopists, irrespective of the type of microscope used. We propose to grade the leukocytes in a different way than searching for leukocyte dominance over epithelial cells, namely by counting them per high power field and per epithelial cell.