The aim of this study was to emphasize the importance of non-surgical treatment for subgroups of patients with spinal epidural abscesses (SEA). From 1988 to 2000, thirty cases of epidural spinal abscesses were retrospectively included in the study. The records and radiological studies were evaluated. Staphylococcus aureus was the most frequent microorganism causing SEA in 18 patients. In 20 patients SEA was secondary to interventional procedures. Predisposing factors were present in 15 cases. Fifty per cent was located in the lumbar region. A total of 22 patients received MRI, which always gave the diagnosis of SEA. Surgical treatment was performed in 20 patients. Conservative treatment with antibiotics was used in 10 patients. Eight patients did not have neurological deficits. One patient was critically ill and another patient was paralysed with an epidural lesion extending over six spinal segments. In all cases, a microorganism was known at the time of diagnosis of ESA. The eight patients without deficits recovered completely following treatment with antibiotics. C-reactive protein was the most reliable inflammatory marker to monitor the effect of the treatment. MRI enables diagnosis of ESA before deficits occur. These can safely be treated with antibiotics if the causative microorganism is known, and the neurological status and laboratory values are monitored. Decompressive surgery is restricted to cases with progressive deficits, when the deficits have lasted for less than 36 h and when the microorganism is not known.