BACKGROUND: For transfusion purposes, blood donors must be accepted both in clinical and serological evaluations and must not have excluded their own donation using the confidential unit exclusion. AIMS: The objective of this study was to verify whether blood donors who choose self exclusion are more likely to be positive in serological tests than donors who do not. METHODS: A cross-sectional analysis was carried out of 51,861 consecutive whole blood donations from January 2004 to December 2008 at a public blood bank in Londrina, Southern Brazil. RESULTS: Self exclusion was chosen in 1672 (3.2%) donations, most frequently by first-time blood donors (p-value < 0.0001), by blood donors from external collections (p-value < 0.0001), by men (p value < 0.0001) and by under 30-year-old donors (p-value < 0.0001). The frequency of positive serology was 5.3% in the group that chose self exclusion and 3.5% in the group that did not choose self exclusion (p-value < 0.0001). CONCLUSIONS: These results show that confidential unit exclusion used in this blood bank is effective and is inexpensive. However, the diagnostic power to detect blood-borne infections was low and resulted in the discard of a high number of blood bags without any direct or indirect serologic markers of pathogens. The use of confidential unit exclusion could be replaced with molecular tests to screen blood donors.