BACKGROUND: High burden of high risk human papillomavirus (HR HPV) has been shown to be predictive for the development of high grade cervical lesions and invasive cancers. However, low viral load cannot inevitably exclude progression towards cervical diseases. Moreover, few studies addressed whether viral load could predict infection clearance. OBJECTIVES: We carried out a retrospective study to analyze the variations of HPV16 load over time as a predictive marker of clinical outcome. STUDY DESIGN: The population consisted of 38 women who were found HR HPV positive by HCII test at study entry. Among them, 13 had developed a CIN2/3 (cases) and 25 had a negative HCII test and a normal cytology (controls) at study exit. The HPV16 DNA loads were quantified in 132 longitudinal cervical samples using quantitative real-time PCR. RESULTS: At study entry, the median of HPV16 load was not statistically different between controls and cases. However, when using a cut-off value of 200 copies/10(3) cells, the rate of cumulative incidence of CIN2/3 at 18 months increased from 14% in women with a load<or=200 copies/10(3) cells to 48% in women with a load>200 copies/10(3) cells. The longitudinal analysis performed on follow-up samples showed that in cases the progression to CIN2/3 was linked to HPV16 burden increasing over time, whereas in controls a decrease of at least 1 log HPV16 DNA load was observed over>or=2 time points. CONCLUSIONS: These results show that kinetics of HPV load, rather than a single HPV detection, might be more reliable to estimate whether a HPV infection will progress or be cleared.