The best method for objectively measuring the level of running speech has not yet been established. In the present study, subjective and objective measures of the level of sentence pairs were compared. Psychoacoustic judgments of the loudness of ten sentence pairs were made by ten listeners. The level of the sentences was measured using four objective methods. The purpose was to determine which of the objective measures best matched the psychoacoustic data. The objective methods included: Sound level meter (SLM), percentile level, average power level (Leq) and the loudness calculation procedure according to ISO 532 B. There were no significant differences between the estimates using the SLM set on either 'Impulse', 'Fast' or 'Slow' settings, the Leq, the percentile level Lmax and the percentile loudness Nmax to N30 using conventional calculation techniques. The measures obtained using A-weighting on the SLM were not found to be superior to the unweighted measures. These findings have implications for the measurement of the level of sentences used in clinical audiology.