Over the last decade, there has been an increase of disease emergence and in particular of vector borne diseases. In order to support specific surveillance or control programmes there is a need for syndromic surveillance. One way to achieve this is to utilise an extensively large amount of data which is produced frequently and automatically by veterinary health agencies, administration bodies and/or farmers’ organisations. The objectives of this study were to develop and evaluate both a set of indicators based on reproduction data, and various statistical methods, to detect the emergence of a disease with known effects on reproduction in dairy cattle. The bluetongue outbreak in four districts in France in 2007 was used as a case model. Five reproductive indicators based on occurrence of return-to-service or gestation length were proposed and evaluated. Using the indicator with the highest elevations during the epidemic, five statistical models combined with a CUSUM (cumulative sum of differences between expected and observed data) method to trigger alarms were implemented and evaluated. Four indicators were able to flag the epidemic in all infected districts except one in which the cumulated yearly prevalence was very low. The indicator based on short gestation had the longest and the earliest elevations from prediction. The alarm system was effective in detecting the epidemic with all statistical models. Performance of the alarm system was good and varied with the threshold parameter value of the CUSUM method. A good compromise between a low number of false alarms (2-4 alarms over 4 years) and early timeliness (~3 weeks) was obtained for a parameter around 0.015.