Allergen immunotherapy (IT) has encouraging therapeutic outcomes but its safety is still being questioned because of possible severe systemic reactions. The aim of this study was to determine the frequency of systemic reactions (SR), and to identify their correlation with the characteristics of therapy, such as allergen composition and IT schedule, and diagnosis. We analyzed the data of 126 patients who received IT between 2000-2003, and suffered from respiratory allergy or hymenoptera venom anaphylaxis. IT was given by rush, clustered or conventional schedules. The standardized allergen extracts used were grass pollen, house dust mite and hymenoptera venom in 88, 18 and 20 patients, respectively. None of the patients received premedication. A total 4705 injections were administered. One hundred and twenty-three adverse events (AE) (2.6% per injection) were documented in 46 patients. Sixty-one of them were SRs (1.3% SRs per injection) and they were seen in 28 patients. Asthmatics had more tendency to SRs (p=0.05). Rush (1.8%) and clustered (2.8%) IT protocols were associated with a higher rate of SRs (per injection) when compared to conventional schedule (0.9%) (rush vs conventional; p=0.013, clustered vs conventional; p=0.001). The majority of SRs corresponded to grade 3 (49%). Forty-nine (80%) of the 61 SRs were observed during the build-up phase, and mostly with pollen extracts (75.5%). Patients showed more severe SRs during the build-up phase (p<0.05). Twenty-six (42.6%) of the SRs were immediate, whereas 35 (57.4%) SRs appeared within 2 hours. Delayed SRs were significantly more frequent in polysensitized patients when compared to monosensitized subjects (p=0.018). Our data indicate that rapid IT regimens and the presence of asthma represent a greater risk for SR development. Since the late SRs occur as frequently as the early ones, we suggest a longer waiting period beyond 30 minutes, especially in polysensitized and asthmatic patients.