Background In spite of many benefits for individuals and community, self-medication has a number of potential risks. Objective To identify predictors of self-medication with over-the-counter and prescription (Rx) medicines without doctor's prescription. Setting Serbian population of 15 years of age and over. Method This was crosssectional, epidemiological study. Data have been drawn from Serbian National Health Survey 2013 database. Predictors of self-medication were determined among sociodemographic, health related and health services related factors, using multivariate logistic regression models. Main outcome measure Prevalence and predictors of selfmedication. Results The study included 14,623 participants. The prevalence of selfmedication was 27.1 and 24% after exclusion of vitamins, minerals and herbal preparations. Medicines for pain relief were the mostly used medicines without doctors' prescription in 18.4% of participants. A relatively high prevalence of self-medication with Rx medicines, (1) tranquilizers and sleeping pills, (2) antibiotics and (3) antihypertensives, was found, 4, 2.5 and 1.9%, respectively. Socio-demographic factors (middle age, female gender, higher level of education), health related factors (chronic disease, stress, physical pain), and health service related factors (nonrealized healthcare needs due to long waiting, dissatisfaction with publicly funded health services) have been found as significantly associated with self-medication. Conclusion Self-medication in Serbia is predicted by socio-demographic and health related factors, unmet needs for healthcare and dissatisfaction with publicly funded healthcare services. Improvements in healthcare system, particularly, shortening of long waiting for healthcare services and improvements in pharmaceutical services, particularly better control of Rx medicines dispensing, could contribute in improvement of responsible self-medication.