Soon after A-weighing began, an article of 1938 mentioned that as a metric it didn't represent true low-frequency loudness. In 1952, a method for “guessing” low-frequency energy was proposed and later, Beranek in 1954 recommended that if the dBC level is greater than dBA, an analysis should be performed below 150 Hz. In 1969, Botsford introduced a criterion for predicting the low-frequency level by [(dBC-dBA) > 10 dB], using the entire audible spectrum, so that this condition applies only if the spectrum is “balanced.” The German Standard DIN 45680 of 1997 proposed a similar criterion, limiting this calculation to one-third octave band frequencies from 8 to 100 Hz. Since 1999, the WHO recommends using the Botsford criterion, but warns that it is not valid if the spectrum contains tones or if it is unbalanced. This paper will demonstrate that the correct way to analyze the low-frequency sound levels is to use its own bandwidth, that to have a traceable noise descriptor the ISO low-frequency definition must be used, it means to use the one-third octave band frequencies between 16 and 200 Hz, to calculate the difference (dBC-dBA) in this range to predict its actual low-frequency sound level.