BACKGROUND Recurrent melioidosis occurs in approximately 6% of patients in the first year following the initial presentation. A recent study revealed that 25% of patients with recurrence had reinfection rather than a relapse resulting from a failure to cure. The aim of this study was to reevaluate these 2 patient groups to define their individual risk factors. METHODS All adult patients who presented to Sappasithiprasong Hospital (Ubon Ratchathani, in northeast Thailand) with culture-confirmed melioidosis during the period 1986-2004 and who survived to receive oral antimicrobial therapy were observed until July 2005. Clinical factors and antimicrobial treatment of patients with recurrent disease due to relapse or reinfection, as confirmed by bacterial genotyping, were compared using a time-varying Cox proportional hazard model. RESULTS Of 889 patients who survived and underwent follow-up, 86 patients (9.7%) presented with relapse, and 30 patients (3.4%) became reinfected. There was no difference in acute outcome between the relapse and reinfection groups. No risk factors for reinfection were identified. Multivariate analyses identified choice and duration of oral antimicrobial therapy as the most important determinants of relapse, followed by positive blood culture result (hazard ratio [HR], 1.86; 95% confidence interval [CI], 1.18-2.92) and multifocal distribution (HR, 1.95; 95% CI, 1.03-3.67). Patients treated with an appropriate oral antibiotic regimen for 12-16 weeks had a 90% decreased risk of relapse (HR, 0.10; 95% CI, 0.02-0.44), compared with patients who were treated for < or = 8 weeks. Trimethoprim-sulfamethoxazole plus doxycycline was an effective oral therapy. CONCLUSIONS This study highlights clinical factors associated with an increased likelihood of relapse and provides evidence for optimal oral antimicrobial therapy.