One contributing factor to the development of nocebo effects is information provided about possible side effects. However, nondisclosure of information can be problematic. We assessed whether positively framed side effect information (highlighting likelihood of not experiencing side effects) can reduce nocebo effects compared to negatively framed information (highlighting likelihood of experiencing side effects). One hundred twelve participants took part in research ostensibly assessing the influence of benzodiazepines (actually sham capsules) on anxiety. Participants were randomized to receive a sham capsule with positively or negatively framed information about four side effects, or a no-treatment control condition. Side effect expectations were assessed after information provision. Framed side effects and other unmentioned symptoms were assessed during the session and 24-hr follow-up. Nocebo effects occurred in symptoms presented as side effects (regardless of framing) during the study session and follow-up (ps < .003). At follow-up, there was also a nocebo effect in other unmentioned symptoms (p = .018). Positive framing reduced side effect symptoms compared with negative framing during the study session (p = .037), but this effect was no longer present at follow-up (p = .53). Side effect expectations did not differ between the framing conditions (p = .14). Positive framing reduced side effects short-term, but not at follow-up. Expectations did not differ between negative and positive framing. Nocebo effects appeared to generalize to other unmentioned symptoms over a 24-hr period. Further research is needed to determine whether the initial impact of positive framing can be maintained over time.