The evaluation of measurement uncertainty of natural gas properties calculated from composition data is an essential aspect of fiscal metering in the trade of natural gas. In particular in gas allocation, it is essential to have a reliable value for the uncertainty. This need is also reflected in the current edition of ISO 6976, the standard for computing natural gas properties, which follows much more closely the requirements of the “Guide to the expression of uncertainty in measurement”. A critical aspect in the uncertainty evaluation of natural gas properties is the fact that the amount fractions of the components of natural gas are always correlated after normalisation. A novel algorithm is provided to recover these correlations from the normalised fractions and associated standard uncertainties. The standard uncertainties are reproduced within 2 , which is acceptable for uncertainty calculations. The correlation coefficients obtained from the recovery algorithm agree with the ones obtained by normalisation. The algorithm requires only the data commonly stored by the natural gas industry as input and it enables providing values and credible associated uncertainties for natural gas properties calculated from composition.