Objectives: To evaluate the incidence and risk factors for liver enzyme elevations (LEE) in patients initiating first-line ART in the ICONA prospective observational cohort, between June 2009 and December 2017. Patients and methods: In total, 6575 ART-naive patients were selected, initiating two NRTIs with the third drug being a boosted PI (n=2436; 37.0%), an NNRTI (n=2384; 36.3%) or an integrase strand transfer inhibitor (INSTI) (n=1755; 26.7%). HBV surface antigen and HCV RNA were detected in 3.9% and 5.8% of the study population. Inverse probability weighted Cox regression analysis was used to calculate the HRs, according to first-line regimen, for LEE, defined as ALT or AST increases of ≥2.5× upper limit of normal (ULN) for patients with normal baseline values or ≥2.5× baseline for patients with higher baseline values. Results: One hundred and eighty-three LEE occurred over 20722 patient-years of follow-up. After adjusting for the main confounders, the risk of LEE halved with INSTIs compared with NNRTIs (HR 0.46, 95% CI 0.25-0.86), with a significant reduction in the raltegravir group (HR 0.11, 95% CI 0.02-0.84 using the NNRTI class as reference). HRs for LEE were significantly higher in subjects with HBV or HCV coinfection, in patients with poorly controlled HIV infection and in those who acquired HIV through homosexual transmission. Conclusions: In our study, INSTI use almost halved the risk of LEE compared with other regimens. This finding could be particularly important for choosing ART in patients with risk factors for liver toxicity such as HCV and HBV coinfections.