The election of the correct loading control in Northern blot normalization is something essential to obtain valid results. Housekeeping genes are widely used as loading control and the assumption is made that they counteract load differences between samples. We have found, however, that uneven sample load is capable to alter the results despite normalization, considering no influence of the experimental conditions on housekeeping gene regulation takes place. Normalization ratio (transcript of interest/housekeeping gene) is determined as the pattern of variation in the ratio between densitometric signals of transcripts--both target and control--and the amount of total RNA. The fact that this relationship is specific for each transcript means different ratios will exist depending on the chosen control gene. Actually, loading differences of only 2 microg may induce a 2.5-fold difference between normalized ratios, depending on the housekeeping gene selected for normalization. In order to select the appropriate loading control, it becomes essential to establish a standard curve for each transcript of interest and several housekeeping. Only the one yielding a constant ratio of normalization along the total RNA range used is to be taken into consideration.