This paper evaluates the effect of herd size on the sera-prevalence of antibodies against Salmonella enterica based on data from the nation-wide salmonella control program (Mousing et al. 1997) taking into account the effects of date of slaughter (season) and geographical county of the herd. Since, there were multiple measurements from each herd, we needed to account for the dependence between samples from the same herd. This led us to construct a random-effects model which allowed interpretation of the size of the fixed effects in relation to the size of the variance of the random effects, that is to quantify the effect of the known variables (included as fixed effects) in relation to the variation in those not known (modelled as random effects). This means that the focus of the statistical modelling will be moved from significance testing of effects to estimation of effects and interpretation of the importance of these relative to each other and relative to the subject matter.