We analyzed the peer review process in the Croatian Medical Journal (CMJ) from 1992 to 1996 by a retrospective analysis of review forms for 319 manuscripts. The forms asked about manuscript's structure (7 questions), its scientific value (7-item scale), clarity and length, and final recommendation (5-item scale). An international manuscript had at least one author affiliated with a non-Croatian institution. The overall rejection rate of manuscripts was 23.5%. National and international manuscripts had similar rejection rates except for original research manuscripts in clinical sciences (34.7% vs. 18.9%, p=0.046). Out-door peer review was asked for 77.4% of the manuscripts; other manuscripts were commissioned and passed an in-house review. Over the years, the number of international reviewers increased, and that of national reviewers decreased. National reviewers more often did not fill in the review form, and international reviewers more often asked for a major revision. The agreement between reviewers ranged from 34.1% (scientific value) to 90.7% (reference citations). Kappa for inter-rater agreement was poor to fair, without difference between national and international manuscripts. International manuscripts had shorter median review time (from receipt to decision) and publishing time (from acceptance to publication) than national manuscripts: 58 vs. 112 days (p<0.001), and 116 vs. 140 days (p<0.009), respectively. Our analysis shows that peer review can be introduced and sustained in a small journal from the scientific periphery. It can be fair both to national and international manuscripts, although work with the authors of national manuscripts may be needed to improve the quality of data presentation.