We describe simple modifications to the ICON II hCG (URINE) pregnancy test to provide a sensitive and specific urinary assay for hCG in field studies of fetal loss. The modified assay had a qualitative lower limit of detection of 0.30 IU/l, a 50% qualitative limit of 0.61 IU/l, a 100% qualitative limit of 1.16 IU/l, and a quantitative limit of 0.80 IU/l. Coefficients of variation ranged from 9.9% to 21.1%. Parallelism was observed among serially diluted subject samples. We used the assay in an 11-month prospective study of fetal loss in rural Bangladesh in which urine samples were collected twice-weekly from 494 women; 330 pregnancies and 93 fetal losses were detected. The median time to a positive pregnancy diagnosis was day 26 from last menses. The modified assay provided qualitative detection of early pregnancy comparable to laboratory assays, and appears to be well suited for use in epidemiologic or rural-population fetal loss studies.