Even apparently subtle disturbance to habitat may have severe long-term consequences if that disturbance alters specific microhabitat features upon which animals depend. For example, in south-eastern Australia, the endangered broad-headed snake Hoplocephalus bungaroides and its prey (velvet geckos Oedura lesueurii) shelter in narrow crevices beneath sun-warmed rocks. Humans frequently displace rocks while searching for snakes and lizards, and these reptiles are rarely found under such displaced rocks (even when the rocks superficially appear suitable). We quantified disturbance to rock outcrops and show that most disturbance was subtle (rocks were typically displaced <30 cm from their original position), but that disturbed rocks harbored fewer reptiles than undisturbed rocks. In a field experiment, we replaced half of the rocks back to their original positions to test whether crevice structure and microclimates differed between disturbed and restored rocks. Crevices beneath displaced rocks were larger and cooler than those beneath restored rocks, and precise repositioning of rocks enhanced usage by reptiles. Both crevice size and temperature influence reptile retreat-site selection; hence, minor displacement of overlying rocks reduces habitat quality by modifying critical crevice attributes. The subtlety of this disturbance suggests that even well-intentioned researchers could damage habitat during field surveys. Conservation of rock outcrop systems requires efforts to reduce rock disturbance, and to educate those searching for animals beneath rocks about the importance of replacing rocks properly. Encouragingly, if rocks are not completely removed, disturbed outcrops can be quickly and easily restored by returning displaced rocks to their original locations.