OBJECTIVE: To report a single-center experience of compassionate use of amphotericin B lipid complex (ABLC) in patients with proven or suspected fungal infection who were or would have been unable to tolerate conventional amphotericin B. METHODS: Twenty-eight patients receiving 30 courses of ABLC for 22 proven invasive mycosis episodes (11 aspergillosis, seven candidosis, four miscellaneous) and eight suspected episodes are described. Seven patients were given ABLC first-line therapy because of conditions precluding the use of amphotericin B deoxycholate (Am B). Twenty-one patients, initially given Am B, were shifted to ABLC because of failure in four, nephrotoxicity of AM B alone or in combination with another drug in 15, and acute side effects in two. The initial dose of ABLC was 5 mg/kg per day; this could be lowered to 3 mg/kg per day or transiently interrupted in cases of impairment of renal function. RESULTS: A mean cumulative dose of 6107 mg (660--16 050) was given over a mean duration of 22 days (4--49). Clinical response rate was 63% (14/22), with mycologic eradication in 37% (9/17) in proven infections. For proven aspergillosis, corresponding rates were 54% (6/11) and 20% (2/10), and in proven candidosis 71% (5/7) and 60% (3/5), respectively. Twenty-one courses were complicated by one or more side effects: fever and chills (11), impairment of renal function requiring a transient reduction of drug dosage (14), hypotension (1). However, for the whole group, creatinine clearance before and after 2, 4 and 6 weeks of treatment remained quite stable. CONCLUSIONS: ABLC, with its low toxicity, enabled us to treat patients who were or would have been unable to tolerate an efficacious dose of Am B. No conclusions about efficacy can be drawn from this small-size, compassionate study. Well-designed studies to compare efficacy and safety of conventional amphotericin B and the various lipidic formulations should be implemented.