An African horse sickness (AHS) outbreak occurred in South Africa's AHS controlled area in autumn 2016. A freedom from disease survey was performed to establish the likelihood of ongoing circulation of the associated virus during the same period the following year. A single‐stage surveillance strategy was employed with a population‐level design prevalence of 1% to establish a survey population sensitivity of 95% (probability that one or more positive horses would be detected if AHS was present at a prevalence greater than or equal to the design prevalence). In March 2017, a total of 262 randomly selected horses from 51 herds were sampled from the 2016 outbreak containment zone. Three within‐herd and herd‐level design prevalence scenarios were used in evaluating the post‐survey probability of freedom. Depending on the underlying design prevalence scenarios, effectively ranging between 0.8% and 6.4%, and the use of informed or uninformed priors, the probability of freedom derived from this surveillance ranged between 73.1% and 99.9% (uninformed prior) and between 96.6% and 100% (informed prior). Based on the results, the authors conclude that it is unlikely that the 2016 AHS virus was still circulating in the autumn of 2017 in the 2016 outbreak containment zone. The ability to perform freedom from disease surveys, and also to include risk‐based methods, in the AHS controlled area of South Africa is influenced by the changing underlying population at risk and the high level of vaccination coverage in the horse population. Ongoing census post‐outbreak must be undertaken to maintain a valid sampling frame for future surveillance activity. The seasonality of AHS, the restricted AHS vaccination period and the inability to easily differentiate infected from vaccinated animals by laboratory testing impact the ability to perform a freedom from disease survey for AHS in the 12 months following an outbreak in the controlled area.