Objectives:This study assessed the ability ofin vitrosusceptibility testing of clinical Candidaisolates to predictin vivoresponse to itraconazole cyclodextrin solution.Methods:One hundred specimens were obtained from HIV-positive patients with oral thrush, of which 72 speciments were from patients who were clinically unresponsive to fluconazole at standard doses and had fluconazole-resistant isolatesin vitro. Susceptibility to itraconazole was assessed by measuring the relative growth of an isolate in liquid medium containing a single concentration of itraconazole and then expressing growth in itraconazole as a percentage of growth in antifungal-free medium.Results and conclusions:Where specimens yielded only one isolate, a cut-off relative growth in itraconazole of 68% discriminated between isolates from patients failing to respond clinically to itraconazole solution and those from patients successfully treated with the preparation (specificity 100%; sensitivity 88%). The presence of mixed infection reduced the predictive accuracy of the test. Only 30% of fluconazole-resistant isolates were cross-resistant to itraconazole. No isolates were resistant to itraconazole but susceptible to fluconazole. Non-response to itraconazole solution was attributed to resistant yeast infection in the majority of cases, and this susceptibility method accurately identified specimens from patients unlikely to respond to the drug.