Abstract Objective To compare the quality of cervicovaginal samples obtained from basic health units (BHUs) of the Unified Health System (SUS) and those obtained fromprivate clinics to screen precursor lesions of cervical cancer. Methods It was an intervention study whose investigated variables were: adequacy of the samples; presence of epithelia in the samples, and cytopathological results. A total of 940 forms containing the analysis of the biological samples were examined: 470 forms of women attended at BHUs of the SUS and 470 forms of women examined in private clinics in January and February of 2016. Results All the unsatisfactory samples were collected at BHUs and corresponded to 4% of the total in this sector (p < 0.0001). There was a higher percentage of samples containing only squamous cells in the SUS (43.9%). There was squamocolumnar junction (SJC) representativeness in 82.1% of the samples from the private clinics (p < 0.0001). Regarding negative results for intraepithelial lesions and/or malignancies, the percentages obtained were 95.9% and 99.1% (p < 0.0049) in the exams collected in the private system and SUS, respectively. Less serious lesions corresponded to 0.89% of the samples from the SUS and 2.56% of the tests from the private sector; more serious lesions were not represented in the samples obtained from BHUs, whereas the percentage was 1.49% in private institutions. Conclusion Unsatisfactory cervical samples were observed only in exams performed at the SUS. There is a need for guidance and training of professionals who perform this procedure to achieve higher reliability in the results and more safety for women who undergo this preventive test.