The measurement manuals published by the Japanese Ministry of the Environment admits use of a sound level meter (class 2), but the manuals requests more precision than the tolerance set by The Measurement Act of Japan. In this study, I examined the bias and dispersion in the measurements obtained by sound level meters in use and discussed how their measurement accuracy varies. And I found secular sensitivity changes in sound level meters occur gradually over the years or suddenly, though such changes are often unnoticeable, as they occur within the tolerance specified by The Measurement Act. This leads to the conclusion that we can efficiently ensure the accuracy of sound level meters by understanding the secular changes and measurement variations, forecasting sensitivity changes from the obtained data, and adjusting the sensitivity during official approval checks in accordance with The Measurement Act. We also examined measurements of sound level meters statistically and noticed that, if we think that all devices tested by The Measurement Act secure the same accuracy, we may assess similar measurements as being significantly different. This may occur because the measurement accuracy within the tolerance often varies by device.