This paper investigates wage differentials between immigrants and natives in Italy along the entire wage distribution and try to account for them using information on observed characteristic of both populations. Analyses are based on data from the 2009 survey on "Income and Living Conditions of Households with Foreigners", the first nation-wide survey aimed at obtaining information on the socio-economic conditions of the foreign population living in Italy, and on the Italian sample of EU-SILC. Immigrants wage gap is disentangled according to three main dimensions: gender, immigrants length of stay in Italy and Italian language proficiency. We show the existence in the Italian labour market of a large wage differential between immigrants and natives which increases along the wage distribution suggesting the existence of a "glass ceiling effect" for immigrants workers. Moreover, we find evidence of (i) a large gender difference in the observed gap, with women showing a higher gap than men, (ii) an ongoing but largely incomplete as similation process among the immigrant population and (iii) a lower gap for immigrants with higher language proficiency. Quantile regressions, Blinder-Oaxaca and Machado-Mata decompositions are applied. The counterfactual decomposition shows that individual characteristics account for only a small fraction of total observed differentials, sector controls increase the explained portion but do not account for the "glass ceiling" pattern of the gap, which instead disappears with occupation controls, suggesting that the upward-sloping shape of the (log) wage differential is a symptom of the occupational segregation characterizing the immigrant population in the Italian labour market.