Outdoor recreation is a valuable ecosystem service permitted in most protected areas globally. Land-use planners and managers are often responsible for providing access to natural areas for recreation while avoiding environmental impacts such as declines of threatened species. Since recreation can have harmful effects on biodiversity, reliable information about protected-area visitation patterns is vital for managers. Our goal was to quantify recreational use in a fragmented urban reserve network and identify factors that influenced visitation. We empirically measured visitation rates at 18 reserves in San Diego County, California. Using random forest models, we identified biophysical and socioeconomic factors that influenced spatial variation in visitation rates and made projections to 27 additional reserves, validating with an expert opinion survey. Visitation rates varied widely across the reserve network. Accessibility variables, such as numbers of housing units and parking lots, were key explanatory variables that had positive relationships with visitation rates. To illustrate the applications of our models, we assessed the exposure of 7 species and subspecies of conservation concern to recreation by comparing predicted occurrence to projected visitation intensities. We found that several species and subspecies, including the orange-throated whiptail (Aspidoscelis hyperythra), western spadefoot (Spea hammondii), and the federally-threatened coastal California gnatcatcher (Polioptila californica californica), are likely exposed to high levels of recreational activity. Our results can be used to identify species for further research, highlight areas with potential conflict between recreation and conservation objectives, and forecast future changes in visitation.