Background. Plasmodium knowlesi commonly causes severe malaria in Malaysian Borneo, with high case-fatality rates reported. We compared risk, spectrum, and outcome of severe disease from P. knowlesi, Plasmodium falciparum , and Plasmodium vivax and outcomes following introduction of protocols for early referral and intravenous artesunate for all severe malaria. Methods. From September 2010 to October 2011 we prospectively assessed nonpregnant patients aged 12 years admitted to Queen Elizabeth Hospital (QEH), Sabah, with polymerase chain reaction–confirmed Plasmodium monoinfection. Standardized referral and prereferral intravenous artesunate were instituted at district hospitals. Results. Severe malaria occurred in 38 of 130 (29%) patients with P. knowlesi , 13 of 122 (11%) with P. falciparum , and 7 of 43 (16%) with P. vivax . The commonest severity criteria in knowlesi malaria included parasitemia >100 000/µL (n = 18), jaundice (n = 20), respiratory distress (n = 14), hypotension (n = 13), and acute kidney injury (n = 9). On multivariate analysis, P. knowlesi was associated with a 2.96-fold (95% confidence interval, 1.19–7.38-fold) greater risk of severity than P. falciparum ( P = .020); only parasitemia and schizontemia >10% independently predicted knowlesi severity. Risk of severe knowlesi malaria increased 11-fold with parasitemia >20 000/µL, and 28-fold with parasitemia >100 000/µL. Nearly all (92%) knowlesi malaria patients received oral artemisinin therapy; 36 of 38 (95%) and 39 of 92 (42%) with severe and nonsevere disease, respectively, also received 1 dose of intravenous artesunate. No deaths occurred from any species. Conclusions. Plasmodium knowlesi is the commonest cause of severe malaria at QEH, with parasitemia the major risk factor for severity. Early referral and treatment with artesunate was highly effective for severe malaria from all species and associated with zero mortality.