Feline leukaemia virus (FeLV) infection in domestic cats can vary in its outcome (persistent, transient, no infection) for reasons that are not entirely known. It was hypothesized that the initial virus and provirus load could significantly influence the course of retrovirus infection. To determine the role of provirus loads, two methods of PCR, a nested PCR and a fluorogenic probe-based (TaqMan) real-time quantitative PCR, which were specific to the U3 region of FeLV-A were established. FeLV provirus in naturally and experimentally infected cats was then measured. Only 3 weeks after experimental FeLV-A infection, persistently infected cats demonstrated higher provirus loads and lower humoral immune responses than cats that had overcome antigenaemia. Lower initial provirus loads were associated with successful humoral immune responses. Unexpectedly, provirus in the buffy-coat cells of two cats that tested negative for the p27 antigen (a marker for viraemia) was also detected. In 597 Swiss cats, comparison of p27 antigen levels with PCR results revealed broad agreement. However, similar to the experimental situation, a significant number of animals (10%) was negative for the p27 antigen and FeLV-positive by PCR. These cats had a mean provirus load 300-fold lower than that of animals testing positive for the p27 antigen. In conclusion, an association between the provirus load and the outcome of FeLV infection was found. Detection of provirus carriers should contribute to further the control of FeLV. In addition, quantification of provirus loads will lead to a better understanding of FeLV pathogenesis and anti-retrovirus protective mechanisms.