ABSTRACT Sources of fecal contamination in two shellfish-bearing tidal rivers (Coan and Little Wicomico) adjacent to the lower Chesapeake Bay in Virginia were identified using antibiotic resistance analysis (ARA) and pulsed-field gel electrophoresis (PFGE) on Escherichia coli. Both rivers contained numerous shellfish beds that were closed to harvest as a result of fecal contamination. Host-origin fecal samples (n = 519) were collected and used to build a regional library for each method (8 E. coli isolates per sample for ARA, 4,152 total; 4 isolates per sample for PFGE, 2,076 total). The best classification rule for the host-origin data set was generated by discriminate analysis (nonpooled covariance, jackknife estimates, 80% threshold), and the average rate of correct classification (ARCC) for the libraries was 70.4% for ARA and 83.3% for PFGE. The libraries were challenged for predictive ability with sets of isolates that represented reproducibility, accuracy, and robustness. The libraries were used to ...