Purpose: The aim of the paper is to investigate earnings management practices related to goodwill accounting, focusing on its first recognition as well as its write-offs, due to the impairment test. Design/methodology/approach: The study refers to a sample of Italian listed firms and the analysis covers three years, with a total of 591 firm-year observations. The modified Jones’ regression model has been used in estimating discretionary accruals, as a proxy of earnings management practices. Findings: A positive relationship between discretionary accruals and yearly changes in goodwill has been proved. Findings also show an incidence of leverage and performance. Research limitations/implications: The study focuses on a single context (Italy) and it is essentially based on financial-economic variables. Practical implications: Findings of the study could be relevant for standard-setters in future revisions of goodwill accounting. Social implication: The study could support investors in evaluating the incidence of first recognition as well as goodwill impairment on the quality of earnings.