This study examined how the species richness and activity patterns of mammals differed in previously unsurveyed habitat fragments with varying recreation intensities on the Palos Verdes Peninsula, Los Angeles County, California. I hypothesized that the number of mammal species detected and activity level of mammals was reduced in habitat fragments with recreation present. Eight game cameras were installed in two habitat fragments in the Palos Verdes Peninsula (PVP), one canyon with recreation (hiking, biking, dog walking) and one reference canyon with minimal human use. Cameras were placed at natural constrictions along the two main travel corridors in the canyons, trails and creek beds. The camera installations were monitored for 200 d. Species richness, activity location, and activity time were analyzed for each canyon. The reference canyon had a higher number of mammal species detected (nine versus six) and accounted for 93% of wildlife detections. Canis latrans showed no difference in the number of detections between canyons, while the other eight mammal species detected exhibited higher activity in the reference canyon. Canis latrans exhibited greater variability in time of activity in the reference canyon than conspecifics in the recreation canyon. Mesopredators Procyon lotor and Felis catus accounted for 67% of wildlife detections in the canyon free of recreation and pose a potential risk to songbirds of conservation concern present in the PVP. Although limited by sampling constraints, this study provides a foundation to investigate wildlife dynamics in the network of habitat fragments in the Palos Verdes Peninsula.