The UK is probably unique in that much information concerning the natural incidence of bee virus diseases in apparently healthy honey bee colonies exists for the period before the arrival of Varroa destructor in 1992 (Bailey et al., 1981). Fig. 1 represents data from 73 individual honey bee colonies infested with V. destructor maintained in Devon or Hertfordshire, UK, between 1993 and 1997 as part of studies investigating the interaction between honey bees, V. destructor and viruses. All colonies were observed for at least 12 months, which included an overwintering period, and were not treated with an acaricide. Natural mite mortality was assessed on a monthly basis using open mesh floors (Carreck, 1994), and the maximum mite population estimated by use of a computer model (Martin, 1998). Samples of dead workers were collected monthly from the majority of the study colonies (others were sampled less frequently) and were tested serologically for the presence of: acute bee paralysis virus; bee virus X; bee virus Y; chronic bee paralysis virus (CBPV); cloudy wing virus (CWV); deformed wing virus (DWV); filamentous virus, Kashmir bee virus (KBV); sacbrood virus (SBV); and slow paralysis virus (SPV), as described previously (Bailey and Ball, 1991). The serological methods have since been largely superseded by molecular techniques such as PCR, but their relative insensitivity actually makes them particularly suited to this type of study, since only infections likely to have caused the mortality of the individual bee would have produced a positive result. Whilst of these 10 viruses, all but KBV were sporadically detected in our samples, only three virus infections, CWV, SPV and DWV were found to be more prevalent in colonies infested with V. destructor in our study compared to previous long term observations (Bailey et al., 1981; Bailey and Ball, unpublished observations) of uninfested colonies in Britain prior to the arrival of the mite. Of these viruses, SPV and DWV are clearly associated with the loss of infested colonies (Carreck et al., 2005; Martin, 2001), but any association between CWV and losses associated with V. destructor remains circumstantial (Carreck et al., 2010). The results showed that colonies in which none of these three key viruses (CWV, SPV, DWV) were detected could survive from one season to the next, even when mite populations exceeded several thousand (Fig. 1). Total estimated mite populations in this group of colonies varied from 60 to 15,000. In contrast, mite populations in the group of colonies that also survived but in which one or more of these three viruses was detected, were uniformly small and did not exceed 2000 mites (Fig. 1). By far the largest group of colonies, however, failed to survive from one season to the next. One or more of the three viruses were detected in the dead adult bees from all of these colonies, particularly DWV and SPV; 14 colonies had one virus, 34 had two, and eight had all three viruses. There was a large variation in the size of the mite population, but almost all colonies had mite populations that exceeded 2000 (Fig. 1). Since the different viruses have different natural histories and epidemiologies, it would seem reasonable that each virus should required different numbers of mites to kill colonies. Both modelling (Martin, 2001) and field observations (Ball, Carreck and Martin, unpublished observations) suggest, perhaps counter intuitively, that a virulent virus such as SPV requires a larger mite population to kill a honey bee colony than a virus such as DWV, which is not rapidly fatal (Sumpter and Martin, 2004). In practice however, few of the colonies studied contained a single virus; most had two, and some had three viruses present. The virus free colonies in our study all occurred within the first few years of V. destructor being found in the UK in 1992, and subsequent studies showed that after a few years, the majority of UK colonies were infected with DWV (Carreck et al., 1999; Ball, 2001). Furthermore, invasion of mites from other colonies is not predictable, so the two colonies in Fig. 1 which died with maximum mite populations of only 984 and 1680, were both in apiaries with other heavily infested colonies which had died. Entry of many mites carrying a virus into relatively mite free colonies may cause their death with fewer mites present than might be predicted from models (e.g. Martin, 2001) which rely on the natural development of the mite population. The results from this study emphasise the continuing need for beekeepers to maintain mite populations at a negligible level by appropriate treatment.