Background-matching is one of the most common strategies for visual camouflage. Its efficacy can be attributed to predator selection for the most conspicuous prey individuals, resulting in the population’s colour patterns resembling the backgrounds within their environment. However, in heterogeneous environments, backgrounds can vary considerably particularly in colour and complexity. This in turn could influence the selection for varied colour and patterns in a prey population to match the different background types. In order to determine if the colour brightness (luminance) and pattern complexity observed in a prey population was selected for background-matching, we conducted a field experiment to compare the frequency of predator attacks on colour pattern variants of replica shore skinks (Oligosoma smithi) in two habitat types at a sand dune system (light sand versus dark vegetation). We created four types of clay models (plain light, plain dark, simple dark, complex dark) and painted them according to reflectance and patterns observed in the population. We expected models that were dissimilar to their backgrounds (e.g., plain light models in vegetation) to be attacked more frequently than ones that matched their background. A total of 240 models were placed in a grid transect on sand (<20% vegetation cover) and vegetated (>80% vegetation) areas in the dunes. Our results show that in the vegetated area, patterned models were attacked more frequently than plain light or plain dark models. There were very few attacks observed in the sand area and thus, no difference between the model types. These results are contrary to our background-matching hypothesis. This may reflect predators’ preference for the most common or abundant colour pattern variant in the population (i.e., patterned individuals) and suggests that predation risk may be influenced more by predators’ search image rather than conspicuousness per se.