Aims: To investigate whether an immunoglobulin (Ig)G avidity ELISA can be used to differentiate between acute and persistent infection with Salmonella (S.) Dublin in cattle. To determine whether the IgG isotype, IgG(1) and IgG(2) responses in acute and persistent infections differ. Methods and Results: Animals were selected from two herds with long-term infection (years) and two herds recently infected (< 3 months). Forty-seven animals were categorized into groups based on the persistence of their antibody level in milk. Based on titre from two serial dilutions the avidity index (AI) was calculated for IgG (IgG-AI), IgG(1) (IgG(1)-AI) and IgG(2) (IgG(2)-AI). The mean IgG-AI for suspected carrier animals with either persistently high (group 1) or persistently high to medium high (group 2) antibody levels was significantly (P = 0.003) higher (32.1% and 38.4%) than for acutely infected animals (21.7% and 22.3%). The probability of being a suspect carrier was associated with IgG-AI, antibody level in the sample and age. However, the effect of age could be the result of a biased sample selection. Specificities and sensitivities were calculated at a range of cut-off values for IgG-AI and IgG(1)-AI. Overall, IgG(2)-AI was high compared with IgG(1)-AI, and there was no difference in IgG(2)-AI between infection groups. There was no difference in the ratio IgG(2) : IgG(1) for acute and persistent infection groups. Conclusions: Assuming that a persistently high antibody response is indicative of persistent infection with S. Dublin in cattle, it can be concluded that the IgG-AI can aid in differentiating between acute and long-term infection on herd level. However, for the test to be useful as an alternative tool to repeated sampling over time for detection of persistently infected carriers during control strategies in cattle herds, the test needs to be optimized and studied further in a larger sample of well-characterized infections in cattle. The affinity of IgG(2) is higher than IgG(1) early in the S. Dublin infection. There appears to be no difference in the IgG(2)-AI between the acute and chronic infection stages. Significance and Impact of the Study: For decades the strategies for detection of persistently infected cattle in S. Dublin infected herds have involved repeated bacteriological culture of faecal samples or repeated antibody measurements over several months. Both methods are time consuming and costly, leaving a new method for detection of carrier animals based on a single sampling highly desirable. This study illustrates a tool, IgG-AI, which may prove useful, although more validation of the method is required before it is used in practice.