This paper examines the reliability of regional democratic consumer price indices. The focus on the production of regional weights and the standard error of regional measures of price inflation arising from uncertainty in the regional weights. Issues in the use of modelling techniques applied to expenditure shares are discussed and it is shown that fixed effects methods may be preferable to random effects techniques. The use of small area techniques with mean log household income used as a covariate is compared with purely data-based measures, and it is found that the use of a covariate does not generally result in an increased level of accuracy.