Ultrasound (US) guidance is frequently used in critically ill patients for central venous catheter (CVC) insertion. The effect of US on infectious risk remains controversial, and randomized controlled trials (RCTs) have assessed mainly noninfectious complications. This study assessed infectious risk associated with catheters inserted with US guidance vs use of anatomical landmarks. We used individual data from 3 large RCTs for which a prospective, high-quality data collection was performed. Adult patients were recruited in various intensive care units (ICUs) in France as soon as they required short-term CVC insertion. We applied marginal Cox models with inverse probability weighting to estimate the effect of US-guided insertion on catheter-related bloodstream infections (CRBSIs, primary outcome) and major catheter-related infections (MCRIs, secondary outcome).We also evaluated insertion site colonization at catheter removal. Our post hoc analysis included 4636 patients and 5502 catheters inserted in 2088 jugular, 1733 femoral, and 1681 subclavian veins, in 19 ICUs. US guidance was used for 2147 catheter insertions. Among jugular and femoral CVCs and after weighting, we found an association between US and CRBSI (hazard ratio [HR], 2.21 [95% confidence interval {CI}, 1.17-4.16]; P = .014) and between US and MCRI (HR, 1.55 [95% CI, 1.01-2.38]; P = .045). Catheter insertion site colonization at removal was more common in the US-guided group (P = .0045) among jugular and femoral CVCs in situ for ≤7 days (n = 606). In prospectively collected data in which catheters were not randomized to insertion by US or anatomical landmarks, US guidance was associated with increased risk of infection.