Abstract Ecological restoration enjoys widespread use as a technique to mitigate for environmental damage. Success of a restoration project often is evaluated on the basis of plant cover only. Recovery of a native arthropod fauna is also important to achieve conservation goals. I sampled arthropod communities by pitfall trapping in undisturbed, disturbed, and restored coastal sage scrub habitats in southern California. I evaluated arthropod community composition, diversity, and abundance using summary statistics, cluster analysis, and detrended correspondence analysis (DCA) and investigated influence of vegetation on arthropod communities with multiple regression analysis. Arthropod diversity at undisturbed and disturbed sites was greater than at sites that were 5 and 15 years following restoration ( p < 0.05). Number of arthropod species was not significantly different among undisturbed, disturbed, and restored sites, and two restoration sites had significantly more individuals than other sites. Vegetation at disturbed and undisturbed sites differed significantly; older restorations did not differ significantly from undisturbed sites in diversity, percent cover, or structural complexity. In multiple regression models, arthropod species richness and diversity was negatively related to vegetation height but positively related to structural complexity at intermediate heights. Exotic arthropod species were negatively associated with overall arthropod diversity, with abundance of the earwig Forficula auricularia best predicting diversity at comparison (not restored) sites (r2 = 0.29), and abundance of the spider Dysdera crocata and the ant Linepithema humile predicting diversity at all sites combined (r2 = 0.48). Native scavengers were less abundant at restored sites than all other sites and, with a notable exception, native predators were less abundant as well. DCA of all species separated restored sites from all other sites on the first axis, which was highly correlated with arthropod diversity and exotic arthropod species abundance. Lower taxonomic levels showed similar but weaker patterns, with example families not discriminating between site histories. Vegetation characteristics did not differ significantly between the newly restored site and disturbed sites, or between mature restoration sites and undisturbed sites. In contrast, arthropod communities at all restored sites were, as a group, significantly different from both disturbed and undisturbed sites. As found in other studies of other restoration sites, arthropod communities are less diverse and have altered guild structure. If restoration is to be successful as compensatory mitigation, restoration success standards must be expanded to include arthropods.