BACKGROUND: Despite the use of closed system drug transfer devices (CSTD), residual contamination from antineoplastic drugs is still detected inside isolators. The aim of this study was to compare the decontamination level obtained using a CSTD + standard cleaning procedure with a CSTD + standard cleaning procedure + specific decontamination procedure. METHODS AND FINDINGS: A comparative and prospective study was carried out in a newly opened compounding unit. Compounding was performed with a CSTD (BD-Phaseal, Becton-Dickinson). In the Control isolator (C), the cleaning process was completed daily with a standard biocide solution (AnioxysprayTM, Anios, France). In the Intervention isolator (I), weekly decontamination with a homemade admixture of sodium dodecyl sulfate 10-2 M/70% isopropanol (80/20, v/v) was added. Monitoring was performed via a validated LC-MS/MS method. Eight drugs (cyclophosphamide, cytarabine, dacarbazine, fluorouracile, gemcitabine, ifosfamide, irinotecan and methotrexate) were monitored daily over 14 consecutive weeks on three sites inside the isolators: gloves, workbench and window. Results are presented as the odds-ratio (OR) of contamination and as overall decontamination efficiency (EffQ, %). The proportion of EffQ ≥ 90% was assessed by a Fisher's exact test (p<0.05). Overall contamination rates (CR, %) were significantly different from one isolator to the other (CRC = 25.3% vs. CRI = 10.4%; OR = 0.341; p<0.0001). Overall EffQ values (median; 1st and 3rd quartiles) were higher in the intervention isolator (I: 78.3% [34.6%;92.6%] vs. C: 59.5% [-5.5%;72.6%]; p = 0.0015) as well as the proportion of days with an EffQ ≥ 90% (I: 42.9% vs. C: 7.1%; p = 0.077) but very variable depending on drugs. CONCLUSION: Adding a decontamination protocol with a tensioactive agent to a CSTD leads to better control of chemical contamination inside isolators. Improving decontamination by increasing decontamination frequency or modifying the protocol will be further studied.