Diagnosis of vaginal discharge is frequently performed in an empirical way, leading to inadequate treatment. This study tested the accuracy of a self-collection kit for microbiological study of the vaginal content. One hundred and forty-two women of Family Health Program units in Niterói and Piraí cities were enrolled in order to have their vaginal content studied. A brief explanation and a self-collection kit were provided in order to sample the vaginal content. The self-collection kit was composed of one empty plastic tube, two glass slides, a long handle cytobrush, an identification card and guideline notes. The vaginal sample was applied on the glass slides by the women and stained by Gram technique. A second sampling was done by the medical personnel. The microbiological diagnosis in a blinded analysis was made under optical microscopy. A validation diagnosis test was done taking the medical collection results as a gold standard. A total of 106 women had followed the protocol and were included in the study. Microbiological analysis was unsatisfactory in 12 cases (6 cases of self-collection material and 6 cases of medical collection). The microbiological analyses in the self-collection and in the medical collection material were respectively: bacterial vaginosis in 21.7% and 17.9%, non bacillar flora in 10.3% and 11.3%, vaginal trichomoniasis in 5.66% and 5.6%, candidiasis in 3.78% and 2.8% and a normal microbiota in 52.8% and 56.6%. The Kappa coefficient suggested a "very good correlation" of the microbiological results between the two methods of collection (K=0.7945). The self-collection kit provides samples for microbiological analysis of the vaginal microbiota as good as medical collection.