There are two main sources of data on income distribution. Household based surveys report mainly on inequality in equivalised household level disposable income. Top income shares, on the other hand, focus on the tax unit as the unit of analysis, because administrative records are obtained from such units. Tax return data is typically analysed in terms of unequvalised fiscal income and obtains better coverage of those at the very top of the income distribution. In this paper, we find that differences in concepts and measures play a very substantial role in accounting for the divergence in the pictures of inequality arising from household and tax return data. Estimates of the share of the top 10% of tax units in fiscal income from the two sources are quite close. Average incomes for the top 1% of the population appear to be substantially higher in tax return data than in SILC - a pattern that has often been observed internationally. We conclude that there is a strong case for examining potential adjustments to survey data to ensure better representation of income levels at the very top of the income distribution.