BACKGROUND: The role of coagulase-negative staphylococci (CNS) in bacteremias continues to be controversial. Until the 1970s, CNS were mostly recognized as contaminants, being part of the cutaneous flora. Since then, several studies have reported increasing incidence and severity of infections due to CNS. PURPOSE: To review the literature concerning the epidemiology of CNS bacteremia in the United States and Europe with reference to the multiple definitions of infection versus contamination, considering the effect of potential biases influencing the validity of the reported results. METHODS: Literature search of the MEDLINE database from January 1980 to February 1998. Studies with fewer than 500 episodes of bloodstream infections or fewer than 100 episodes of CNS bacteremia were not included in the pooled analysis. RESULTS: (1) CNS remain the most frequent contaminants (58%-83% of positive blood cultures); (2) the proportion of all bloodstream infections caused by CNS is increasing (R=.51); (3) the overall incidence of true CNS bacteremia is increasing (R=.54, P=.0014); (4) comparing the United States to Europe, there is an increasing trend in the incidence of nosocomial bacteremia due to CNS in the United States (R=.82, P=.0006), but no trend is seen in European studies; (5) the mortality associated with true CNS bacteremia varies between 4.9% and 28%. DISCUSSION: This review confirms the increasing importance of CNS bacteremias, measured both as a proportion and as an incidence of bloodstream infections. The contributions of several possible explanations for the incidence increase and the difference between the United States and Europe need further evaluation: (1) increased recognition and awareness of CNS infections among clinicians; (2) a gradual change in the definition of true bacteremia from an obligatory two positive blood cultures to one positive blood culture associated with a clinical picture compatible with infection; (3) a change in blood culture practices and techniques; (4) an increase in the numbers of blood cultures performed, which is reported both in the United States and in Europe; (5) a shift toward more elderly patients with increasingly severe underlying illnesses; and (6) increasing use of intravascular devices. CONCLUSIONS: The apparent trend of increasing CNS bacteremia seems to be valid. Whether there is a real difference between the United States and Europe concerning the increase of CNS bacteremia is difficult to establish due to the large number of confounding factors. Few studies take into account the number of blood cultures performed or the use of intravascular devices to adjust for the observed trends. Further on-site surveillance studies are needed to investigate the phenomenon more extensively.