As epidemiologists, we agree with the chief scientist of the UK Department for Environment, Food and Rural Affairs that badger culling is an “evidence-based” policy for controlling bovine tuberculosis (TB) in Britain (Nature http://doi.org/mxd; 2013). We disagree with other, less positive interpretations of that evidence (Nature 490, 317–318; 2012).

The evidence comes from a large-scale, long-term, government project — the Randomised Badger Culling Trial (RBCT). This roughly halved the incidence of TB in cattle herds in the culling area following 4–7 annual badger culls (H. E. Jenkins et al. PLoS ONE 5, e9090; 2010). The reductions in disease were not immediately evident owing to the long incubation period of bovine TB. There were also transient increases in incidence outside the cull areas, attributed to increased badger movements.

The mixed reaction to the RBCT results might reflect the way the data were collated and interpreted. Widely quoted summary statistics from the RBCT include data collected at the start (before any effect was likely) and long after culling had ceased. In our view, this approach is appropriate for testing for statistical significance but is not the best way to quantify effectiveness for a trial.

These analyses were never intended to account for the nonlinear impact of sustained reductions in transmission. For bovine TB in England, where various control measures are already in place, basic epidemiological principles predict that even a small reduction in transmission rate could have large benefits.

All this has confused the debate, and risks compromising the interpretation of future trials of alternative control options, such as badger vaccination.

Planned pilot studies will establish whether badger control is practicable, humane and safe. It is for society to decide whether badgers should be culled at all.