We describe a system for multiattribute drug product searching. We then demonstrate the system's performance on sample queries, and evaluate the name-based similarity searching component. Ten drug names were used to query a database of existing drug names using five different retrieval methods. Retrieved names were merged into master lists and presented to 15 pharmacists. Pharmacists rated the similarity between the query name and each retrieved names on a scale of 1-5. We report the precision of our five different retrieval methods at 11 levels of recall. The best single measure was editex, with a precision of 17.4% averaged across 11 levels of recall. A regression model using four objective measures of similarity as predictors accounted for 40.6% of the variance in observed mean similarity ratings. Automated, multiattribute drug product searching may improve the effectiveness and efficiency of preapproval screening processes and thereby prevent medication errors.