In a prospective randomized open study, ceftriaxone, 50 mg/kg per day, was compared with ampicillin, 100 mg/kg per day, both given for a period of 5 days, for the treatment of 40 children whose mean (+/- SD) age was 4.5 +/- 3.2 years and who had severe dysentery caused by Shigella organisms. Twenty patients were treated with ceftriaxone and 20 with ampicillin. Both drugs were initially given intravenously for a period of 1 to 2 days and were continued intramuscularly, in the case of ceftriaxone, or orally, in the patients receiving ampicillin. All Shigella organisms isolated were susceptible to ceftriaxone; 28% were resistant to ampicillin. The diarrhea persisted for a mean (+/- SD) period of 2.5 +/- 1.0 days in the ceftriaxone-treated patients versus 6.8 +/- 6.3 days in the ampicillin-treated patients (p less than 0.005). At the end of the 5 days of therapy, stool cultures for Shigella organisms were negative in 12 (60%) of the 20 patients from the ampicillin group and in all the children (100%) from the ceftriaxone group (p less than 0.001). Bacteriologic relapses were observed in eight (40%) of the patients treated with ampicillin but in none of the children treated with ceftriaxone (p less than 0.001). In instances of clinical or bacteriologic failure in the ampicillin group, retreatment was instituted in most of the cases with ceftriaxone; persistent clearing of the Shigella organisms from the stool was finally achieved after a mean (+/- SD) period of 11.75 +/- 9.4 days after therapy was started, as compared with 1.85 +/- 0.6 days in the ceftriaxone-treated patients (p less than 0.001). We conclude that in children with severe shigellosis, treatment with ceftriaxone for 5 days is effective and better than use of ampicillin for clinical cure and eradication of the Shigella organisms from the stool.