This study tested whether a redesign methodology would improve the quality of medical records in the obstetric services of four Ecuadorian hospitals. Quality teams in each hospital implemented a redesign methodology, working in cooperation with local quality assurance (QA) experts from the QA Project and following a predetermined sequence of steps. Eight quality standards for medical records were defined: (1) complete set of forms, (2) correct chart headers, (3) complete discharge summary, (4) complete delivery form, (5) patient consent, (6) identification number on admission and discharge forms, (7) legibility, and (8) coherency and consistency. Pre- and post-samples of medical records (448 before and 459 after) were audited to determine compliance with the eight standards. The average increase in the eight indicators for the four-hospital pooled sample was 27 percentage points, up from 41 percent compliant in the pre-sample to 68 percent in the post-sample. Five of the indicators showed highly significant gains of 25 percentage points or more, with four of them attaining post-intervention compliance of 80 percent or more. Across the four hospitals, pre-intervention average compliance ranged from 27 to 49 percent; the average gain ranged from 24 to 31 percentage points. The gain at each hospital was statistically significant, but the differences among the hospitals were not. A secondary purpose of the study was to test and improve the redesign methodology. The study was carried out sequentially, one hospital at a time: participants made recommendations for improving the methodology, those recommendations were used to modify the methodology, and the modified version was used at the next hospital. In effect, although small modifications were made to the redesign methodology as the study progressed, there was no evidence that these modifications improved compliance with the standards. Prior to redesign, the quality of the obstetric medical records was very poor, well under 50 percent compliance. Such poor documentation is not suitable for use in quality assessment or for proper management of patients. The systematic and participatory redesign methodology applied in this study was very successful in increasing the quality of the medical records, especially for the indicators of completeness, legibility, and coherency, but less so for indicators related to patient signature and patient identification number. Future research is needed to test whether the improved quality of these records is adequate to monitor changes in the quality of obstetric care. (author's)