The aims of this study were to evaluate six portable blood glucose meters for use in cats in a clinical setting and to identify potential sources of inaccuracy such as the effect of glucose concentration and haematocrit. Excess fluorinated whole-blood samples were obtained and were tested using the six meters and a reference laboratory method. Bland-Altman plots were constructed and an error grid analysis was performed, using a grid adapted in this study for diabetic cats. Error grids are a clinically oriented non-parametric approach to blood glucose data, and are designed to determine whether differences between glucometer and laboratory readings are clinically significant. All the meters studied had the potential to under- or overestimate blood glucose levels to varying degrees throughout the glycaemic range. This variation was not consistent enough to be predictable and correctable. The study failed to demonstrate any link between haematocrit and difference between laboratory and glucometer readings.