Phase variation of type 1 pili (fimbriae) was studied during the in vivo growth of Escherichia coli in two animal models. In the first, a heavily piliated urinary tract isolate (strain 149) was placed in 1-cm polypropylene chambers sealed with 0.22-μm-pore-size filters. The chambers were surgically implanted intraperitoneally in mice and recovered at various times. Piliation, as determined by electron microscopy and by measuring the minimum number of bacteria needed to produce mannose-sensitive hemagglutination, gradually decreased, and by day 5, most of the organisms were nonpiliated. In the second model, piliated and nonpiliated E. coli phase variants were inoculated into the bladders of BALB/c mice via urinary catheters, and their fate in the lower urinary tract was studied. Viable counts of bladder homogenates revealed that piliated phase variants were significantly more effective in colonizing the bladder urothelium than were their nonpiliated counterparts. Specific antibody to type 1 pili prevented colonization by the piliated organisms. After inoculation of piliated variants, the bladder-associated bacteria gave rise to approximately 80% mannose-sensitive hemagglutination-positive colonies, and immunocytochemistry of bladder lavages revealed large numbers of type 1 piliated bacteria adhering to the bladder transitional cells. Electron microscopy confirmed the presence of piliated bacteria in association with the bladder urothelium. The urine of these mice, whose bladders were colonized with piliated bacteria, frequently showed no growth, and when bacteria were present, strain 149 yielded less than 30% hemaglutination-positive colonies. The results suggest that for some E. coli strains, phase variation may be a factor in determining the fate of the E. coli in the urinary tract and that the urine may not necessarily reflect the bacteriologic state of the bladder mucosa.