The paper estimates the extent of evasion of personal income tax (PIT) in Italy by integrating two methods that the literature has previously applied separately. The consumption-based method introduced by Pissarides and Weber (1989) is used to estimate underreporting of income in micro data collected in the household IT-SILC survey. We adopt an econometric specification close in spirit to that of Feldman and Slemrod (2007), which allows us to estimate income underreporting at different rates for different income sources. The underreporting estimates are then used in the discrepancy method to correct the incomes declared by the households in the survey and to compare them with administrative data. The comparison provides new estimates of evasion of personal income tax by type of income, region and income class. The estimates are used to improve microsimulation analyses of the distribution impact of tax evasion