Objective: to assess the degree of reproducibility in the reading of cervical cytology smears among four specialized reading centers at Medellin, Colombia. Methodology: 181 cervical cytology smears from a study on the prevalence of cytological abnormalities in Pueblorrico, a town located in the southwestern region of the state of Antioquia in Colombia, were re-read by four specialized centers in Medellin. Each center was asked to conduct a blind routine reading of the smears to avoid disclosure of information between centers. The reproducibility was measured by percentage agreement and kappa. Results: of 181 smears only 55 matched across all 4 centers, with an overall concordance rate of 30% and an overall kappa of 0.31. According to Fleiss scale, poor reproducibility was observed. The concordance between pairs of centers ranged between 0.3 y 0.7. Discussion: there is a high variability in the interpretation of cytological results among centers in Medellin. It is necessary to implement training processes and to unify readings for cervical cytology criteria. ----------