Background Unique blood culture (UBC) has been proposed to limit the number of venipuncture and to decrease the risk of BC contaminations (BCC) without affecting their yield. We hypothesized that a multi-faceted program based on UBC in the ICU may reduce the rate of contaminants with a similar performance for bloodstream infections (BSI) identification. Methods In a before and after design, we compared the proportion of BSI and BCC. A first 3-year period with multi-sampling (MS) strategy followed by a 4-month washout period, where staff received education and training for using UBC, and a 32-month period, where UBC was routinely used, while education and feedback were maintained. During the UBC period, a large volume of blood (40 mL) was sampled through a unique venipuncture with additional BC collections discouraged for 48 h. Results Of the 4,491 patients included (35% female patients, mean age 62 years) 17,466 BC were collected. The mean volume of blood per bottle collected increased from 2.8 ± 1.8 mL to 8.2 ± 3.9 mL between the MS and UBC periods, P < 0.01. A 59.6% reduction (95% CI 56.7–62.3; P < 0.001) of BC bottles collected per week was observed between the MS and UBC periods. The rate of BCC per patient decreased between the two periods from 11.2% to 3.8% (73.4% reduction; P < 0.001) for the MS and UBC periods, P < 0.001. Meanwhile, the rate of BSI per patient remained stable at 13.2% and 13.2% for the MS and UBC periods, P = 0.98. Conclusions In ICU patients, a strategy based on UBC reduces the contamination rate of cultures without affecting their yield. Graphical Abstract