Housing unit listing is often used in countries that do not have household or person registries to create frames for household surveys. While several studies have reported the kinds of units and areas that are at risk of overcoverage and undercoverage in such frames, none has looked at variability in the listing process. This article explores this variability by comparing two frames created by trained field staff using the same methods and materials. The overall overlap rate between the two listings is 80%. In nearly all blocks, the listers created different frames, and in more than ten percent of the blocks, the two frames did not overlap at all. In this observational study, the overlap between the two frames is particularly low in the blocks listed using the traditional (from scratch) listing method. There is also evidence that sometimes one lister visited the wrong block. The results show that the listing process can introduce variance into survey data.