Objective: To determine the variability in human papillomavirus (HPV) DNA testing of the cervix in young women found positive for HPV DNA by dot blot hybridization on routine examination. Methods: Young women who were found to be HPV DNA‐positive on routine screening using an RNA‐DNA dot blot hybridization method were asked to return for repeat HPV DNA sampling, cytology, colposcopic examination, and biopsy if indicated. Those who had no histologic evidence of cervical dysplasia were asked to return every 4 months for cytology and HPV DNA testing using standardized RNA‐DNA hybridization and polymerase chain reaction techniques. Results: The women were followed for a mean of 27.6 months (range 13‐40) with a mean of six visits (range four to ten). One‐third of the women remained consistently or intermittently HPV DNA‐positive by RNA‐DNA dot blot hybridization, and almost 50% of the women remained consistently or intermittently positive using polymerase chain reaction techniques. Women were more likely to be positive by polymerase chain reaction than by RNA‐DNA hybridization at both 1 year and after 2 years of follow‐up (P < .05). However, rates for persistent positive tests by either method were similar at 1 and 2 years of follow‐up. Forty percent of the subjects had new or different types than the original HPV type appear during follow‐up. All five women who had evidence of spontaneous regression of cytologic abnormalities became HPV DNA‐negative by both methods. Conclusions: Our data suggest that a portion of women infected with HPV appear to eliminate the infection over a relatively short period and are at low or no risk of developing disease. Persistent DNA negativity was also found in those women undergoing spontaneous regression. However, a substantial proportion of women remained intermittently positive by RNA‐DNA hybridization and polymerase chain reaction. This finding suggests that the virus remains latent in some individuals and may undergo reactivation, defined by sufficient replication to allow detection by means less sensitive than polymerase chain reaction. (Obstet Gynecol 1993;82:578‐85)