A cross-sectional study was conducted in four rural communities of northeastern Trinidad to determine the microbial quality of water supply to households and that quality's relationship to source and storage device. Of the 167 household water samples tested, total coliforms were detected in 132 of the samples (79.0%), fecal coliforms in 102 (61.1%), and E. coli in 111 (66.5%). There were significant differences among the towns in the proportion of the samples contaminated with coliforms (P < 0.001) and E. coli (P < 0.001). Of 253 strains of E. coli studied, 4 (1.6%) were mucoid, 9 (3.6%) were hemolytic, and 37 (14.6%) were nonsorbitol fermenters. Of 69 isolates of E. coli tested, 10 (14.5%) were verocytotoxigenic. Twenty-eight (14.0%) of 200 E. coli isolates tested belonged to enteropathogenic serogroups. Standpipe, the most common water source, was utilized by 57 (34.1%) of the 167 households. Treated water (pipeborne in homes, standpipes, or truckborne) was supplied to 119 households (71.3%), while 48 households (28.7%) used water from untreated sources (rain, river/stream, or well) as their primary water supply. The type of household storage device was associated with coliform contamination. Water stored in drums, barrels, or buckets was more likely to harbor fecal coliforms (74.2% of samples) than was water stored in tanks (53.3% of samples), even after controlling for water source (P = 0.04). Compared with water from other sources, water piped into homes was significantly less likely to be contaminated with total coliforms (56.9% versus 88.8%, P < 0.001) and fecal coliforms (41.2% versus 69.8%, P < 0.01), even when the type of storage device was taken into account. However, fecal contamination was not associated with whether the water came from a treated or untreated source. We concluded that the drinking water in rural communities in Trinidad was grossly unfit for human consumption, due both to contamination of various water sources and during household water storage.