As countries move towards malaria elimination, methods to identify infections among populations who do not seek treatment are required. Reactive case detection, whereby individuals living in close proximity to passively detected cases are screened and treated, is one approach being used by a number of countries including Swaziland. An outstanding issue is establishing the epidemiologically and operationally optimal screening radius around each passively detected index case. Using data collected between December 2009 and June 2012 from reactive case detection (RACD) activities in Swaziland, we evaluated the effect of screening radius and other risk factors on the probability of detecting cases by reactive case detection. Using satellite imagery, we also evaluated the household coverage achieved during reactive case detection. Over the study period, 250 cases triggered RACD, which identified a further 74 cases, showing the value of RACD over passive surveillance alone. Results suggest that the odds of detecting a case within the household of the index case were significantly higher than in neighbouring households (odds ratio (OR) 13, 95% CI 3.1-54.4). Furthermore, cases were more likely to be detected when RACD was conducted within a week of the index presenting at a health facility (OR 8.7, 95% CI 1.1-66.4) and if the index household had not been sprayed with insecticide (OR sprayed vs not sprayed 0.11, 95% CI 0.03-0.46). The large number of households missed during RACD indicates that a 1 km screening radius may be impractical in such resource limited settings such as Swaziland. Future RACD in Swaziland could be made more effective by achieving high coverage amongst individuals located near to index cases and in areas where spraying has not been conducted. As well as allowing the programme to implement RACD more rapidly, this would help to more precisely define the optimal screening radius.