Variable-retention (VR) logging practices provide an alternative to clearcutting, but much uncertainty exists on their effectiveness in maintaining biodiversity. We compared patterns of abundance of terrestrial gastropods in areas subjected to either clearcutting or VR treatments, in relation to an uncut control at six experimental sites in coastal British Columbia before and 2–4 years after logging. Gastropods sensitive to the logging treatments in most comparisons included Haplotrema vancouverense, Pristiloma stearnsii and P. lansingi (as a group), and Striatura pugetensis. Several generalist species showed no response to the treatments, and the abundance of two species (Punctum randolphii and Vespericola columbianus) increased in some logged treatments relative to the control. At sites where trees were retained in small groups (0.2–0.5 ha), the abundance of four species was depressed when compared to the control and pre-logging values. No differences among retention levels of 10%, 20%, and 30% were found. At a site where trees were retained in groups of different sizes, large groups (0.8–1.2 ha) were more effective in supporting sensitive species than were small groups (< 0.2 and 0.2–0.5 ha) and clearcuts. At a site where dispersed trees were retained, none of the logged treatments were equivalent to the control. No consistent patterns of higher abundance in the 30% retention level than in the 5% and 10% levels were found. The results contribute to the growing body of information on the complexity of responses of forest floor organisms to habitat modification by logging.