We studied the prevalence of antibody to Toxoplasma gondii in 883 Costa Ricans between the ages of 15 and 26 years (mean 20.3). There were seven populations, two rural and five urban. The overall antibody prevalence was 61.4%. Because antibody prevalence was 57.1% in the 14 people who ate raw meat, and 61.6% in the 294 people who consumed raw eggs, meat and eggs could be excluded as important modes of transmission. We ascertained the presence of cats, type of soil contact, and flooring in the kitchen for every residence in which each individual lived from birth until interviewed. The 66% of people who indicated contact with cats had an antibody prevalence of 64%, and those without contact 56% (P = 0.02). In six of the seven populations, we found a positive correlation between cat contact and antibody prevalence; hence these six were studied together, and San Ramon, with a negative correlation, was examined separately. In the San Jose metropolitan population and in the total urban sample, cat contact was associated with a significantly higher antibody prevalence (P < 0.01). It was found that humans, even without acknowledged soil contact, acquire infection inside houses, especially with cement floors, when cats defecate in the house. Humans also acquire infection in or around houses with wooden floors, where the crawl space is an often unrecognized refuge of cats and a source of persistent soil contamination. San Ramon was the urban area with the highest human-cat contact (77%) and with the highest antibody prevalence in cats (75%). Although suggesting equal opportunities for infection, irrespective of cat contact, we cannot be certain in a retrospective study in which most of the infections took place an average of 15 years earlier. The most frequent antibody titer in our Costa Rican population was 1:1,024 (21%), much higher than the most frequent titers in a Colombian (1:64 to 1:128) and a U.S. population (1:16) of comparable age.