Introduction Autoantibodies are useful markers for clinical onset of autoimmune diabetes, and as a diagnostic supports. The aim of this study was to determine the prevalence of autoantibodies at the onset diabetes mellitus in our environment, measured by an ELISA technique, and to assess its role as autoimmune diabetes diagnostic support. Material and methods A sample of 111 patients with diabetes onset was studied, which included, 61 type 1 diabetes (12 Latent Autoimmune Diabetes in Adults (LADA) and 50 type 2 diabetes. Control group: 64 non-diabetics. Antibodies glutamic acid decarboxylase antibodies (GADA), membrane phosphatase anti-tyrosine (IA-2), and anti-insulin (IAA) were determined by enzyme immunoassay ELISA microplates. A descriptive observational study was conducted to the measure diagnostic tests using the software PASW Statistics version 18. Results GADA(+): 78.7% DM1; 2.0% DM2; 1.6% control. IA-2(+): 52.2% DM1; 45.5% LADA; 16.4% DM2; 12.5% control. IAA(+): 10.3% DM1; 18.0% LADA; 6.0% DM2; 1.7% control. Comparing means, autoimmune diabetics were significantly differentiated ( P<.0001) by GADA and IA-2 values, but not with IAA. The area under the Curve using software Receiver Operating Characteristics (ROC): GADA=0.90 ( P<.0001), IA-2=0.74 ( P<.0001), IAA=0.57 ( P=.126). Optimal cut off for GADA>5.5U/mL (sensitivity 79%, specificity 98%) and IA-2>6.0U/mL (sensitivity 54%, specificity 84%). On analysing risk factors associated with autoimmune diabetes diagnosis, the calculated Odds ratio gave GADA=51.44 and IA-2=4.64. Conclusions In this study GADA(+) prevalence is high at diabetes onset, and its calculation has been effective for autoimmune diabetes diagnosis. IA-2 increases diagnostic probability 4.6 times. Anti-insulin antibodies measured by ELISA test did not demonstrate value for autoimmune diabetes diagnosis and classification, although its usefulness is not excluded for the diagnosis support of other diseases.