Incidence of seroconversion to caprine arthritis-encephalitis virus (CAEV) was determined for 1,194 goats on 11 dairies, using 2 repeated annual herd tests for CAEV. Current life table methods were used to compare age-specific incidence of seroconversion for pasteurized milk-raised and unpasteurized milk-raised goats. Logistic regression models were used to determine the risk factors associated with CAEV seroconversion, and to estimate odds ratios for seroconversion for various factor levels. Goats raised by unpasteurized milk-feeding methods were 2.5 to 6.7 times more likely to seroconvert than were goats raised by pasteurized milk-feeding methods, depending on the method of comparison. Similarly, 61.6 to 85.0% of seroconversions in yearling goats possibly were attributable to unpasteurized milk feeding. Among yearling goats, CAEV seroconversion was associated with feeding method, breed, and source of goat (herd of origin) when the effect of dairy was considered. In addition to the 6.7 times greater risk of seroconversion for unpasteurized milk-raised goats, yearling goats of the Saanen and Toggenburg breeds were 2.2 and 3.3 times, respectively, more likely to seroconvert than were Alpine yearling goats. Yearling goats purchased from another source were less likely to seroconvert than were yearlings raised on the dairy where they were studied. Among goats > 1 year old, age was associated with risk of seroconversion. Goats that were 3 years old or were > or = 4 years old were 1.7 and 3.2 times, respectively, more likely to seroconvert than were 2-year-old goats, when adjusted for effect of dairy. The effects of dairy were significant (P < or = 0.001) in yearling and older goats.