This paper investigates the effectiveness of predator culling as a means of reducing livestock losses using hunting club data for Cooper (outside Mossel Bay) for the period 1976 to 1981. Results showed that caracal (Caracal caracal) culling increased subsequent livestock losses when compared to farms where fewer caracals were culled. When controlling for lagged rainfall, remoteness and a proxy for other unobserved farm characteristics, a logit model indicated the marginal effect of culling to be a 17.5% increase in the likelihood of livestock losses during the next year. The corresponding negative binomial model estimated the effect of an additional caracal culled to be a 0.373 unit increase in the number of sheep lost. A lagged rainfall variable was negative and significant in both models. According to the logit results, the marginal millimetre of rain reduced subsequent losses by 1.1%. For the negative binomial model, the marginal effect of rainfall was reduced losses by 0.047 of a sheep, which was about a 5% increase in losses. The average number of livestock lost was 0.94 sheep per farm per year. Distance travelled, used to proxy remoteness, was positive in the negative binomial model and non-significant in the logit model. Lagged livestock losses were not significant in either model. This result is important because it provides support for stricter predator control regulations by showing that livestock farmers are inadvertently harming their own interests through inappropriate culling, a practice which continues to this day. Keywords: human-wildlife conflict, culling effectiveness, sheep, caracal, compensatory breeding.