Objective Evaluation of relevance and feasibility of universal newborn congenital cytomegalovirus infection (cCMVI) screening in saliva. Design Retrospective, population‐based cohort study. Setting Clamart, France, 2016–2020. Population All neonates born consecutively in our level III maternity unit. Methods CMV PCR in saliva for all neonates at birth, and, if positive, CMV PCR in urine to confirm or exclude cCMVI. Prospective and retrospective characterisation of maternal infections. ROC curve analysis to assess saliva PCR performances. Acceptability of screening among staff members evaluated by a survey. Main outcome measures Number of cCMVI neonates; number of expected and unexpected cCMVI. Results Among 15 341 tested neonates, 63 had cCMVI (birth prevalence of 0.4%, 95% CI 0.3–0.5). In 50% of cases, maternal infection was a non‐primary infection (NPI) during pregnancy. cCMVI was expected or suspected (maternal primary infection PI, antenatal or neonatal signs) in 24/63 neonates (38%), and unexpected in 39/63 neonates (62%). The best CMV saliva threshold to predict cCMVI was 356 (2.55 log) copies/ml 95% CI 2.52 log–3.18 log, with an area under the ROC curve of 0.97. Over 90% of the 72 surveyed staff members reported that the screening was easy and quick. No parent refused the screening. Conclusions Universal screening for cCMVI with CMV PCR on saliva samples is feasible and highly acceptable to parents and healthcare providers. Over half (62%) of the cases had no prenatal/neonatal signs of cCMVI or a maternal history of CMV infection during pregnancy and would probably not have been diagnosed without universal screening. Tweetable In 62% of congenital cytomegalovirus infection cases, only universal neonatal screening in saliva can detect infection. Tweetable In 62% of congenital cytomegalovirus infection cases, only universal neonatal screening in saliva can detect infection.