Cognitive Bias Modification for anxious interpretations (CBM-I) is a computerized training program designed to reduce threatening interpretations of ambiguous information. In addition to testing the efficacy of CBM-I for contamination fear on various cognitive outcomes and emotional vulnerability, the current study examined whether the content of CBM-I must match a person's fear domain to be efficacious, and examined the moderating role of preexisting interpretation bias on CBM-I's efficacy. Participants (n = 121) high in contamination fear were assigned to one session of either CBM-I focused on contamination fear content, CBM-I focused on height fear content, or a sham control condition. Planned contrasts indicated that fear domain match does tend to enhance CBM-I's effects: compared to the heights condition, those in the contamination training condition improved on five of the eight cognitive bias-related variables, though there was no change on measures of emotional vulnerability. In contrast, there was little difference between the heights training and sham conditions. Minimal evidence was found for baseline interpretation bias as a moderator. Results support the use of CBM-I that matches the fear domain of the participant, regardless of baseline bias level.