The use of linear mixed models (LMM) and its Best Linear Unbiased Prediction (BLUP) methodology is becoming increasingly popular amongst research scientists dealing with data modelling. Development of computer power and user-friendly statistical software facilitate their rapid implementation. This paper gives an overview of the benefits of the flexible LMM/BLUP framework to analyse experimental data collected in sugarcane-breeding programs to make efficient breeding and selection decisions, or to study genetic properties of traits of agronomic interest. Several applications of mixed models are presented. Their common feature is to exploit co-variances existing between some experimental data. Depending on applications, the nature of co-variances differs and might be related either to spatial, temporal and/or genetic considerations, with possible nested effects. In all cases, the methodology has the advantage of providing unbiased statistical inferences and predictions supporting objective conclusions in scientific issues investigated.