Many problems relating to the diagnosis and epidemiology of nosocomial candidiasis remain unsolved. Antemortem diagnosis of invasive candidiasis has always been problematic. Antigen detection systems also appear to be of low sensitivity and specificity. Good epidemiology of nosocomial candidiasis has been handicapped by the relatively poor typing methods available forCandida albicans. Genotyping methods, particularly those based on DNA probes, appear to be major improvements. Using genotyping, two previously described single-strain outbreaks have been reported to be due to multiple-strain types. Nevertheless, studies of nosocomial candidiasis have suggested that cross-infection does occur and can be controlled. There have been several reports from developed countries showing that nosocomial mycobacterial disease still occurs. Nosocomial viral infections remain a major problem, particularly in pediatrie hospitals. These infections may be due to cross-infection from other patients or members of staff. Control measures can be successful if rigorously applied. This includes ensuring that medical staff are adequately vaccinated.