Diabetes mellitus occurs in schizophrenia patients at higher rates than in the general population. Reasons for this elevated risk are poorly understood and have not been examined prospectively in antipsychotic-naïve, first-episode patients. This study aims to determine which antipsychotics are associated with diabetes development in antipsychotic-naïve schizophrenia patients. All antipsychotic-naïve patients diagnosed with schizophrenia in Denmark between 01 January 1997 and 31 December 2004, followed until 31 December 2007, allowing for >or=3 years follow-up, unless death or diabetes onset occurred. Risk factors for the time to diabetes onset were assessed, including antipsychotics taken for at least 180 defined daily doses in the first year after first antipsychotic prescription ('initial treatment'). Risk factors for diabetes incidence were assessed, including antipsychotic use within 3 months before diabetes onset or study end ('current treatment'). Of 7139 patients, followed for 6.6 years (47,297 patient years), 307 developed diabetes (annual incidence rate: 0.65%). Time to diabetes onset was significantly shorter in patients with higher age (hazard ratio (HR): 1.03, confidence interval (CI): 1.02-1.03) and those with 'initial' treatment of olanzapine (HR: 1.41, CI: 1.09-1.83), mid-potency first-generation antipsychotics (FGAs) (HR: 1.60, CI: 1.07-2.39), antihypertensive (HR: 1.87, CI: 1.13-3.09), or lipid-lowering drugs (HR: 4.67, CI: 2.19-10.00). Significant factors associated with diabetes within 3 month of its development included treatment with low-potency FGAs (odds ratio (OR): 1.52, CI: 1.14-2.02), olanzapine (OR: 1.44, CI: 1.98-1.91), and clozapine (OR: 1.67, CI: 1.14-2.46), whereas aripiprazole was associated with lower diabetes risk (OR: 0.51, CI: 0.33-0.80). In addition to general diabetes risk factors, such as age, hypertension, and dyslipidemia, diabetes is promoted in schizophrenia patients by initial and current treatment with olanzapine and mid-potency FGAs, as well as by current treatment with or low-potency first-generation antipsychotics and clozapine, whereas current aripiprazole treatment reduced diabetes risk. Patients discontinuing olanzapine or mid-potency FGA had no increased risk of diabetes compared with patient not treated with the drugs at anytime.