Database on clinically treated patients in Belgrade served as an example for analysis of possibilities for the use of Personal Identifying Number (PIN) as an Unique Patient Identifier. In the first part of the paper we analyzed filling up of the fields which contained data on PIN within complete databases in 1981, 1991 and 1996. Filling up of PIN was significantly changed in the three observed years: it was 18% in 1981; 68% in 1991, and 56% in 1996 respectively. Analysis of interactions among the chosen factors (type of hospital, demographic and social characteristics of patients, length of stay in hospital, manner of treatment, main diagnosis, treatment outcome) and measuring time, showed a different degree of statistical significance. In the second part of our paper we analyzed the unexpected decrease in filling up of PIN in 1996 (as compared to 1991) ussing the method of logistic regression, on 1% samples from the databases for the two respective years. On the basis of obtained models of filling up of UPI data, taken as dependent variable and the above factors (predictors) we analyzed the advantages and drawbacks of UPI application as an unique patient identifier.