The measurement of transcript levels constitutes the foundation of today's molecular genetics. Independent of the techniques used, quantifications are generally normalised using invariant control genes to account for sample handling, loading and experimental variation. All of the widely used control genes are evaluated, dissecting different methodological approaches and issues regarding the experimental context (e.g. development and tissue type). Furthermore, the major sources of error are highlighted when applying these techniques. Finally, different approaches undertaken to assess the invariance of control genes are critically analysed to generate a procedure that will help to discern the best control for novel experiments.