Quality has become the key concept in official statistics. There is a general consensus that we have to consider several components when assessing the quality of statistical information. Relevance, accuracy, timeliness, punctuality, comparability, coherence, accessibility and clarity are the dimensions most frequently mentioned. In this article we use regression analysis to evaluate the contribution of these different dimensions when assessing the overall quality of statistical products. We do this using the information collected in the structured consultation with users and experts from both inside and outside the Spanish Central Administration carried out by the Working Group of the Spanish High Council on Statistics, responsible for the preliminary draft of the proposals and recommendations of this council for the Spanish Multiannual Statistical Programme 2013-2016. We find that the abovementioned dimensions have different weights in the overall assessment of perceived quality (with accuracy and reliability having the highest weight, and relevance having the lowest) and that the structure differs between both types of users. © Statistics Sweden.