Comparing deaths averted by vaccination campaigns is a crucial public health endeavour. Excess all-cause deaths better reflect the impact of the pandemic than COVID-19 deaths. We used a seasonal autoregressive integrated moving average with exogenous factors model to regress daily all-cause deaths on annual trend, seasonality, and environmental temperature in three Italian regions (Lombardy, Marche and Sicily) from 2015 to 2019. The model was used to forecast excess deaths during the vaccinal period (December 2020-October 2022). We used the prevented fraction to estimate excess deaths observed during the vaccinal campaigns, those which would have occurred without vaccination, and those averted by the campaigns. At the end of the vaccinal period, the Lombardy region proceeded with a more intensive COVID-19 vaccination campaign than other regions (on average, 1.82 doses per resident, versus 1.67 and 1.56 in Marche and Sicily, respectively). A higher prevented fraction of all-cause deaths was consistently found in Lombardy (65% avoided deaths, as opposed to 60% and 58% in Marche and Sicily). Nevertheless, because of a lower excess mortality rate found in Lombardy compared to Marche and Sicily (12, 24 and 23 per 10,000 person-years, respectively), a lower rate of averted deaths was observed (22 avoided deaths per 10,000 person-years, versus 36 and 32 in Marche and Sicily). In Lombardy, early and full implementation of adult COVID-19 vaccination was associated with the largest reduction in all-cause deaths compared to Marche and Sicily.