Several prescription procedures for fitting hearing aids use individually measured loudness growth functions or average loudness data to produce a target. Two investigations are reported that suggest that there could be problems using loudness data for hearing aid fitting. In one study loudness growth functions were measured on twenty ears using two different loudness scaling tests (The Contour test and NAL's Adaptive Categorical Loudness Scaling test - NAL-ACLS). The measured data were related to the normative data established for each of the two tests to derive the prescribed compression ratio according to loudness normalisation. Different compression ratios were prescribed in more than half of the cases for which the Contour test consistently prescribed higher ratios than NAL-ACLS. Intra-subject variability could not explain the variation, so which loudness test produces the 'best' compression ratio? In another study normal hearing listeners balanced octave band filtered babble noise to equal loudness. The balancing test was completed with open ears (free field listening) and occluded ears (sounds presented via a transducer and a foam eartip). The data showed that loudness balance judgments obtained with the occluded ear require the level in the ear canal of low frequency stimuli relative to high frequency stimuli to be 5 to 10 dB greater than when the ear is open. If loudness functions are measured with insert earphones and used in hearing aid fitting to normalise loudness, can we be sure that the relative loudness perception of low and high frequency sounds in ears occluded by hearing aids is the same as in the open ears of normal hearing people?