The Messidor database, containing hundreds of eye fundus images, has been publicly distributed since 2008. Created to evaluate automatic lesion segmentation and diabetic retinopathy grading methods, it has been widely used by researchers. However, maintaining and managing such a database is costly. The authors analyze feedback on the database after over six years of distribution, highlighting the importance of such databases for research. The database contains 1200 color fundus images acquired by three ophthalmology departments. It is divided into three sets, each containing 400 images. Each set includes 100 TIFF images and an Excel file with medical diagnoses. Unlike other databases, the Messidor database does not include manual annotations on the images. The database download process requires users to fill in personal and professional information, and the database is hosted by Crihan. The authors note that the number of download requests and website visitors has increased significantly since 2008, with a clear correlation between these metrics and the number of citations in scientific papers. The database has been cited three times more often in 2013 than in 2011, matching the increase in download requests. The authors conclude that the Messidor database has been successful in the field of retinal image processing and diabetic retinopathy diagnosis. However, they note that the images in the database are outdated, as they were acquired before 2007. Despite this, the database remains a valuable resource for researchers. The authors also provide recommendations for the design of future databases, emphasizing the importance of clear descriptions, user acknowledgment, and automatic validation procedures. The study highlights the important role that databases play in medical image processing and the need for new databases that reflect current clinical practices.The Messidor database, containing hundreds of eye fundus images, has been publicly distributed since 2008. Created to evaluate automatic lesion segmentation and diabetic retinopathy grading methods, it has been widely used by researchers. However, maintaining and managing such a database is costly. The authors analyze feedback on the database after over six years of distribution, highlighting the importance of such databases for research. The database contains 1200 color fundus images acquired by three ophthalmology departments. It is divided into three sets, each containing 400 images. Each set includes 100 TIFF images and an Excel file with medical diagnoses. Unlike other databases, the Messidor database does not include manual annotations on the images. The database download process requires users to fill in personal and professional information, and the database is hosted by Crihan. The authors note that the number of download requests and website visitors has increased significantly since 2008, with a clear correlation between these metrics and the number of citations in scientific papers. The database has been cited three times more often in 2013 than in 2011, matching the increase in download requests. The authors conclude that the Messidor database has been successful in the field of retinal image processing and diabetic retinopathy diagnosis. However, they note that the images in the database are outdated, as they were acquired before 2007. Despite this, the database remains a valuable resource for researchers. The authors also provide recommendations for the design of future databases, emphasizing the importance of clear descriptions, user acknowledgment, and automatic validation procedures. The study highlights the important role that databases play in medical image processing and the need for new databases that reflect current clinical practices.