In a retrospective analysis, we assessed the usefulness of two serologic enzyme-linked immunosorbent assays as epidemiologic tools for the detection of cryptosporidiosis episodes in children from a Peruvian community. The incidence rate determined by the serologic assay was higher than the rate determined by stool microscopy (0.77 versus 0.41 infection/child-year of surveillance).Most studies designed to characterize the epidemiology of Cryptosporidium sp. infection have relied on stool microscopy to identify infection episodes. However, because microscopy is relatively insensitive when small numbers of oocysts are excreted (acid-fast stool microscopy is 83.7% sensitive and 98.9% specific relative to PCR) and because the period of oocyst shedding can be short, many infections may escape microscopic detection (2,5,9). Earlier work demonstrated that serum immunoglobulin G antibody responses to two immunodominant sporozoite surface antigens develop upon infection and can be detected for some time after resolution (half-life of 12 weeks) (4,6,7,8). To assess the usefulness of these antibody assays as epidemiologic tools, we assayed longitudinally collected serum samples from children who participated in a birth cohort study of diarrheal disease in Lima, Peru, between 1995 and 1998 (1, 3, 10).Data were assessed using multiple linear and Poisson regression, pooled t tests, chi-square test, and Wilcoxon test. When applicable, the generalized estimating equation procedure was used to adjust for correlation among multiple responses from the same child. Analyses were performed using SAS version 8.0, Sudaan version 8.0.2, or SigmaStat version 2.03.0 (SPSS, Inc.). Statistical significance was set at 0.05.As previously described, stool specimens were collected from study participants at weekly intervals (more frequently when diarrhea occurred or when enteric protozoa were detected) and examined by microscopy for Giardia sp., Cryptosporidium sp., and Cyclospora sp. (1). The current study included all cohort children with more than one microscopyconfirmed cryptosporidiosis episode who donated multiple serum samples (n ϭ 28); all those with one cryptosporidiosis episode who donated Ն7 sera (n ϭ 29); and 17 randomly selected children with no microscopic evidence of Cryptosporidium infection who donated Ն7 sera. Using the infection episode definition of Bern et al. (1), the selected children had 92 Cryptosporidium infections detected by microscopy during 224.3 child-years of stool surveillance (0.41 infection/child-year of follow-up). They were similar to the excluded cohort participants (n ϭ 158) in terms of mean age at enrollment (16 days versus 15 days, respectively; P ϭ 0.67), sex (54% male versus 58% male; P ϭ 0.55), and incidence of diarrhea (6.2 episodes/ year of follow-up versus 7.3 episodes/year of follow-up; P ϭ 0.22) but had more days of follow-up (mean of 1,091 days versus mean of 667 days; P Ͻ 0.0001) and a higher proportion of stools that were positive for Cryptosporidium (0.019 versus 0.010; P ϭ 0.0005), Giardia (...