Several explanations have been proposed to explain these divergent results-failure to define the population being studied; differing, possibly incorrect methods of sample collection; failure to obtain truly "normal" subjects for control purposes, etc. To avoid these various criticisms we have studied a group of normal, asymptomatic women who had not been hospitalized and compared the findings with those in three groups of patients, the pattern of whose illness had been carefully defined by long-term follow-up in a specialist urinary infection clinic. To reduce sampling errors and to define the regular carrier state of these women each was studied weekly for five or six weeks. To exclude the possible influence of recent infection (Marsh et al., 1972) or antibiotic treatment or a changing state before subsequent infection (Stamey et al., 1971) all the patients were known to have had no bacteriuria or antibiotics for six weeks before the study and no bacteriuria during or within six weeks of completion of the study.The results show no difference in the introital carriage rate of E. coli between normal subjects and patients referred with a history of previous urinary infection whether they were subsequently shown to be intermittently bacteriuric, symptomatic and abacteriuric, or asymptomatic and abacteriuric. A small difference in periurethral carriage of enterobacteria in general was observed between normal subjects and the patient groups though there was no difference between the patient groups. Thus it does seem that patients with a previous history of urinary infection may have a slight change in the overall periurethral colonization as compared with normal women but that there is no difference in this respect between those who are subsequently intermittently bacteriuric and those who are persistently abacteriuric. Furthermore, though E. coli was the most common cause of bacteriuria in our patients no difference in the periurethral carriage rate was observed between the normal subjects and the three patient groups. There was also no difference in the recovery rate or persistence of particular E. coli serotypes.When these results are compared with other reports it must be noted that slightly differing techniques are used. Thus Stamey et al. (1971) studied clean-catch urethral urine samples (labelled VBJ) and samples from the vaginal vestibule, Bailey et al. (1973) studied swabs from the external urethral meatus (periurethral swabs) and vaginal vestibule, and Cox et al. (Cox, 1966;Cox et al., 1968) obtained urethral specimens using a specially designed instrument. We have previously used the term "introital" swab to define what is more correctly a periurethral swab. We have also previously shown and subsequently confirmed that colony counts of the same organism are obtained in the same proportion of instances from urethral urine (vB,) and periurethral swabs in normal women and patients alike.By dividing patients into "persistent", "intermittent," and "non-") carriers of introital bacteria on the basis of sw...