Sanitary risk inspection, the systematic observation of contamination hazards, is often used to manage safety of water sources such as wells and boreholes. However, the replicability of sanitary risk inspections undertaken by different observers has not been studied. This study aimed to assess the replicability of sanitary risk inspections by two different observers in urban and peri-urban neighbourhoods of Greater Accra, Ghana. Two observers independently used a standard protocol to record contamination hazards around 62 groundwater sources, additionally recording urban-specific hazards such as damaged sewage pipes via a modified protocol. We calculated risk scores as the proportion of hazards observed at each source, separately for each observer. Linn’s concordance correlation coefficient indicated very high agreement between the two observers’ risk scores (n = 62; c = 0.949, 95% confidence limits 0.917–0.968). However, risk scores from urban-specific observations were uncorrelated with those from the standard protocol (r = 0.11, p = 0.41 for observer 1; r = 0.16, p = 0.22 for observer 2). Ours is the first study of replicability of sanitary risk observations and suggests high inter-observer agreement. However, urban contamination hazards were not captured using the standard protocol. In the future, assessment of inter-observer agreement and observations of urban-specific hazards could be incorporated into nationwide or regional sanitary risk surveys.