OBJECTIVES: To evaluate the performance and cost effectiveness of the WHO recommendations of incorporating risk-assessment scores and population prevalence of Neisseria gonorrhoeae (NG) and Chlamydia trachomatis (CT) into vaginal discharge syndrome (VDS) algorithms. METHODS: Non-pregnant women presenting with VDS were recruited at a non-governmental sexual health clinic in Sofia, Bulgaria. NG and CT were diagnosed by PCR and vaginal infections by microscopy. Risk factors for NG/CT were identified in multivariable analysis. Four algorithms based on different combinations of behavioural factors, clinical findings and vaginal microscopy were developed. Performance of each algorithm was evaluated for detecting vaginal and cervical infections separately. Cost effectiveness was based on cost per patient treated and cost per case correctly treated. Sensitivity analysis explored the influence of NG/CT prevalence on cost effectiveness. RESULTS: 60% (252/420) of women had genital infections, with 9.5% (40/423) having NG/CT. Factors associated with NG/CT included new and multiple sexual partners in the past 3 months, symptomatic partner, childlessness and >or=10 polymorphonuclear cells per field on vaginal microscopy. For NG/CT detection, the algorithm that relied solely on behavioural risk factors was less sensitive but more specific than those that included speculum examination or microscopy but had higher correct-treatment rate and lower over-treatment rates. The cost per true case treated using a combination of risk factors, speculum examination and microscopy was euro 24.08. A halving and tripling of NG/CT prevalence would have approximately the inverse impact on the cost-effectiveness estimates. CONCLUSIONS: Management of NG/CT in Bulgaria was improved by the use of a syndromic approach that included risk scores. Approaches that did not rely on microscopy lost sensitivity but were more cost effective.