We conducted a cross-sectional study of the spatial distribution of HTLV-I/II infection among blood donors of Hemominas Foundation, living in Belo Horizonte, from 1994 to 1996. Study population (1,022) was composed by 533 cases (positive Western Blot (WB), indeterminate WB and ELISA positive without WB result) and a random sample of 489 non-cases (HTLV-I/II serum negative). Cases and non-cases were georeferenced using the exact or an approximation of the household address reported at the blood donation interview. Using multivariate analysis, cases with WB result are less likely to be reposition blood donors compared to voluntary ones (OR = 0.70; CI 95%: 0.50-0.99). Using the difference between univariate K functions, we found no evidence that cases and non-cases differ in their spatial distribution. We found no evidence that cases with and without WB result differ in the distance between their residence and Hemominas Foundation. No donors without WB result were georeferenced by the exact address. These donors could not have received the Hemominas letter inviting them to return to collect the second blood sample.