The incidence of bacteremia following injection sclerotherapy for esophageal varices has been a subject of interest since its resurrection in the early 1980s. Multiple invasions of large venous channels through a needle contaminated with salivary microflora might be expected to be associated with a high frequency of bacteremia, especially in cirrhotic patients with "compromised" immune system. Injection of saliva into peripheral veins has been shown to produce fever and polymicrobial bacteremia.1 In this issue of the Archives, Low and colleagues2 studied 38 patients subjected to 104 sessions of sclerotherapy and compared blood cultures obtained before and after the procedure. Only 4.3% of the postsclerotherapy blood cultures yielded bacteria. My colleagues and I3 previously showed that the incidence of bacteremia following sclerotherapy (5%) is not higher than after routine diagnostic endoscopy (1.8% to 8%).4-6 However, other investigators have reported a higher frequency of bacteremia7,8 and attributed their findings to the length