The role of antifungal prophylaxis remains controversial and concerns exist that the use of azoles may potentiate the emergence of resistant Candida species. We used a strategy of combining the latest azole/triazole with oral amphotericin B to reduce this risk. We analysed data on Candida colonization and candidaemia in neutropenic patients from four prophylaxis periods (1985/6: ketoconazole and amphotericin B suspension; 1991/2 & 1997: fluconazole and amphotericin B suspension; 1998/9: itraconazole) to look for evidence of the emergence of potentially resistant species. Overall, the percentage of patients colonized with Candida fell significantly (69·3%, 57·5%, 43·2% and 46%, respectively, P < 0·001) due to a decrease in colonization with C. albicans (49%, 23·1%, 22·2% and 25·2%, respectively, P < 0·001). However, in 1998/9, increased colonization, particularly with C. glabrata in the lower gastrointestinal tract, was noted to coincide with the omission of oral amphotericin B. Despite an increasing population of ‘high risk’ patients, the incidence of candidaemia has not changed significantly (2%, 1·4%, 1·2% and 2% respectively). However, species causing candidaemia have changed, with resistant organisms now predominating. Our findings support the use of azole prophylaxis although, in view of the trends noted when itraconazole was used alone, we would recommend the additional use of oral amphotericin B.