Abstracts in Inpharma, deHaen Drugs-in-Use, International Pharmaceutical Abstracts (IPA), Reactions, and Clin-Alert were evaluated using the number of omissions and errors as endpoints. Fifty-one abstracts were compared with original articles by pharmacy students using a standardized checklist for drug therapy and adverse drug reaction articles. The overall omission rate per abstract was 3.2 +/- 2.5. There were no significant differences in omission rates in drug therapy abstracts between Inpharma, deHaen, or IPA abstracting services. For adverse drug reaction abstracts, IPA had significantly more omissions than Clin-Alert (p less than 0.001) and Reactions (p less than 0.01). One error occurred in approximately every fifth abstract. No differences were shown between abstracting services in this regard. Although abstracts and abstracting services can provide useful information--particularly when information is needed in a timely fashion, when an original article is difficult to obtain, or when written in a foreign language--they should not be considered a consistently reliable source of information upon which to base decisions, as evidenced by the number and types of omissions and errors described in this study.