Since 1990, unusually high rates of false-positive serological reactions (FPSR) in bovine brucellosis screening have been observed in some countries of the European Union. The aim of this survey was to describe this phenomenon in a highly affected French Department, and to evaluate the links between some individual or herd factors and the occurrence of these FPSR. Before 1990, low backgrounds of FPSR were recorded (individual prevalence rate: less than 0.5 per 10,000). The phenomenon burst during the 1990-91 screening campaign, reached a peak in 1992-93 (50.5 per 10,000), and then decreased until the last studied campaign, 1995-96 (9.1 per 10,000). The phenomenon was transient and sporadic within a herd. At the herd-screening level, four assumed risk factors were isolated: (i) the probability of a herd-screening to be positive was closely and positively linked with the herd screening size; (ii) during a given screening campaign, the prevalence of FPSR decreased from December to November; (iii) the presence of at least one goat on the premises increased the risk for the 1992-93 and 1993-94 screening campaigns; and (iv) a previous FPSR in a given herd appeared to be a weak but significant risk factor. At the individual-animal level, herd size, sex and breed did not seem to be linked with FPSR appearance, while young animals were significantly more affected than older ones. However, global variations in herd or individual prevalences remained unexplained. The lack of link between FPSR and brucellosis is strengthened. The hypothesis of a widely spread causal agent with a low individual host susceptibility and/or a low probability of detecting FPSR animals can be supported by these results.