Point-of-care glucometry is used commonly in clinical and research settings; however, accuracy and precision of this method are concerns. The objectives of this study were to determine the accuracy of glucometry in adult horses and the precision of duplicate measurements. Blood samples were collected from 62 horses into one plain syringe, one EDTA tube and three fluoride oxalate (FO) tubes. Immediately after collection, glucose concentrations in whole blood were determined, in duplicate, by glucometry from the syringe (plain whole blood [WB] group), EDTA tube (EDTA group) and one FO tube (FO group). One FO sample was used to measure plasma glucose concentration by a laboratory chemistry analyser (LAB group) ≤1 h after collection. The third FO tube was used to measure plasma glucose concentration by glucometry after 3 h storage (FO3hr group). Adequate precision was present for all groups (coefficient of variation: 0.7–3.5%) except WB (5.5–9.4%). Between groups, correlations were significant (P < 0.05; except for WB-EDTA), varied with group comparison, and tended to be lowest for comparisons involving WB. Mean bias was lowest for WB-LAB and greatest for FO-LAB and FO3hr-LAB; however, the limits of agreement were ≥4.65 mmol/L for WB-LAB and ≤2.75 mmol/L for most other comparisons. For the glucometer used, performance was influenced by sample type: WB was unsuitable, while FO or EDTA samples resulted in adequate precision and accuracy, provided under-estimation of glucose concentrations is accounted for by using method-specific reference ranges. Glucometer performance and optimal sample type(s) should be determined prior to use in horses.