Salmon lice Lepeophtheirus salmonis Krøyer have caused disease problems in farmed Atlantic salmon Salmo salar L. since the mid-1970s in Norway. High infection intensities and premature return of wild sea trout Salmo trutta L. were first reported in 1992. Later emaciated wild Atlantic salmon smolts carrying large amounts of lice have been observed both in fjords and offshore. The Norwegian Animal Health Authority regulations to control the problem, which came into operation in 1998, included compulsory louse level monitoring in farms and maximum legal numbers of lice per fish. Here, we present a model of salmon louse egg production in Norway and show that the effect of the current public management strategy is critically dependent on the yearly increase in salmon production. This is because the infection pressure is the product of the number of fish in the system, and the number of lice per fish. Due to the much larger number of farmed than wild salmonids, it is highly likely that lice originating from farmed salmon infect wild stock. Estimated tolerance limits for wild salmonids vary widely, and the level of louse egg production in farms which would be needed to decimate wild populations is not known. Two possible thresholds for total lice egg production are investigated: (1) 1986 to 1987 level (i.e. before adverse effects on sea trout were recorded), and (2) a level corresponding to a doubling of the estimated natural infection pressure. The farm lice per fish limits that would have to be observed to keep louse production within the 2 thresholds are calculated for the period 1986 to 2005. A steady decrease in the permitted number of lice per fish may keep the total louse production stable, but the number of salmon required for verification of lice numbers will increase as the prevalence to be verified is decreased. At threshold (2), the model estimated that lice limits should have been 0.05 louse per fish in 1999. This would require 60 fish from each pen to be collected, anaesthetised and examined for a good estimate at a confidence level of 95%. Such sample numbers are likely to be opposed by farmers. The use of national delousing programs to solve the problem is discussed.