BACKGROUND Children attending child care centers have high rates of cytomegalovirus (CMV) excretion. Women exposed to such children have an increased risk of acquiring CMV infection, and primary infection places the offspring of such women at risk of congenital CMV infection. We studied family child care homes to determine if this child care alternative might represent a safe haven from CMV. METHODS One hundred thirty-two women providing care in their homes were studied using a latex agglutination method to determine the rate of CMV seropositivity at baseline. Women who were seronegative for CMV were then sampled prospectively at 6-month intervals between March 1991 and August 1994 to determine the annual rate of CMV acquisition. A point prevalence of CMV excretion in family homes was determined by sampling 106 children from 25 randomly selected homes. Cytomegalovirus isolates were compared by molecular analysis using polymerase chain reaction-based methods to identify transmission. RESULTS At baseline, 57.6% of the 132 providers were seropositive for CMV. Seropositive providers were more likely to be caring for toddlers (aged 1-2 years) (67% vs 46%; P=.02) and had worked in child care somewhat longer (median of 28.5 vs 21.5 months; P=.11). Using stepwise logistic regression, the strongest predictors of seropositivity at baseline were caring for children aged 1 to 2 years (odds ratio [OR] =2.37; P=.02) and number of months as a child care provider (OR= 1.17 for an increase of 24 months as provider; P=.08). Six or more years as a provider was highly associated with seropositivity (OR=3.27; P=.02). During follow-up, 5 of 51 seronegative providers seroconverted, yielding an annual infection rate of 6.8%. The point prevalence survey of children from the 25 homes (14 had seropositive providers) identified 8 CMV-excreting children. Three children in 1 home had indistinguishable isolates by polymerase chain reaction mapping. The provider seroconverted and excreted an isolate with a molecular profile indistinguishable from that of the children. CONCLUSIONS The prevalence of CMV excretion is low among children attending child care homes (8% vs 15% in prior studies of child care centers; P=.07), and only 1 (20%) in 5 of the homes had CMV-excreting children. However, the overall CMV seroconversion rate of home child care providers was comparable to the rate observed among providers in child care centers. Families who use family home child care as an alternative to large child care centers are exposed to a low and unpredictable risk of CMV infection.