Introduction: Planning cluster randomized studies requires special attention due to their specific design. To achieve balance on a cluster level as well as on individual level, it is necessary to apply randomization techniques which involve restricted randomization. Objective: Determine randomization techniques as well as their frequency in protocols for cluster randomized trials. Materials and method: Searching the MEDLINE bibliographic database, there were 1020 bibliographic units, the analysis included only the protocols for cluster randomized trials, which was a total of 169 trials. Data on randomization techniques, units of randomization and publication years of protocols were extracted. Results: The randomization technique with most frequency was stratification (35.9%). After stratification the most frequent was simple randomization (13.5%), followed by a combination of block and stratification (10%), block randomization (9.4%) and matching (9.4%). The most frequent units of randomization were health facilities (52%). The number of published protocols statistically increases during time (p<0.01). Conclusion: The most frequent randomization technique used by researchers is restricted randomization