Until recently, accurate microbiological diagnosis of invasive aspergillosis (IA) was seldom established in HSCT recipients. Blood samples are rarely positive for Aspergillus species, the reliability of the cultures depends of the specimen (if taken from a normally sterile site or not) and biopsy samples require invasive procedures, rarely recommended in patients with severe thrombocytopenia. Implementing the international consensus defining the microbiological criteria for the diagnosis of Aspergillus infection, we retrospectively evaluated the role of serum galactomannan (GM) detection by EIA to diagnose IA among HSCT patients with proven invasive fungal infection (IFI) and the impact of serum storage in GM concentrations. The EIA assay allowed categorizing as "probable" 5 of the 10 cases of "possible" aspergillosis (50%). Considering a lower cut-off level for the reaction (1.0), 80% of the cases could be categorized as "probable" aspergillosis. Positive or undetermined results were detected one to 4 months before the diagnosis of IA in eight of the 11 patients (72.7%) with proven IFI. Retesting the stored samples after a second storage for four years, we could observe lower reactivity in 20% of the samples. The detection of galactomannan by the EIA test represents a major advance in the diagnosis of IA in HSCT recipients at high risk of IA. A better understanding of the kinetics of the GM in different clinical situations is necessary to maximize the benefit of the test in Aspergillus surveillance.