Background: Abstracts of Cochrane reviews are the most used part of reviews. The titles of abstracts are changing from the style ‘intervention for outcome-in condition’ to the style ‘intervention-for condition’. This makes it necessary that the primary outcomes are clearly stated in the abstract. Some abstracts do not clearly present the primary outcomes or present statistically significant results first. Objectives: To estimate the prevalence of problems with reporting primary outcomes. Methods: A regular sample of reviews published in issue 2008-1 of The Cochrane Library was evaluated. Methodological reviews and reviews without studies included were excluded. For 100 reviews, the abstract was compared to the full review. Results: In 58% of abstracts, the primary outcome is not clearly stated in the background, objectives and methods sections. In 17% of abstracts, primary outcome is not the first reported. In these cases, 12 of 17 abstracts presented first the other outcome(s) which demonstrate the statistically significant effect. In 23% of abstracts, it is not clear which of primary outcomes is reported (phrasing is different from review and description is not clear). For primary outcome, the size of effect is not reported in 25% of abstracts, and in 26% the statistical significance of effect is not given. In a total of 70% of abstracts, the effect for the first reported outcome is statistically significant, but in 12/70 it is not the primary outcome. Some statistically not significant findings are presented in 76% of abstracts. Conclusions: Problems in reporting primary outcomes are quite prevalent. They need to be urgently addressed, because Cochrane abstracts are universally accessible and are widely used for informing decision making in time or resource limited situations.