After the SARS-CoV-2 outbreak in spring 2020, Italy faced a second epidemic wave in autumn. Using a SIRD model calibrated on COVID19-related deaths, we describe the regional epidemic dynamics from August to November 2020. We explore the time-varying reproductive number, R0(t), and quantify the number of infections, included their submerged portion, under different infection fatality rate scenarios. Results indicate that during the second epidemic wave, R0(t) changed over time heterogeneously across regions, with some important common elements including a mid-October peak and a decline during November, which suggest the possible role in inflating or deflating the contagion rate of specific events (e.g. schools reopening, regional elections) and of the restrictions imposed at the national and local level to reduce the infection spread. Despite the decline of R0(t) in most regions, the prevalence of circulating infections estimated at the end of the study period was not negligible, in particular in the North of the country. This suggests that even small increases of R0(t) in December may lead in a short time to unsustainable levels of contagion spread, depending on the regional supply of hospital and ICU beds and healthcare services throughout the territory.