Interest for knowing long-term human exposure levels due to mobile communications has increased in the last years. It has been shown that short-term exposure assessment made under standard procedural restrictions is not reliable when it comes to conclusions on long-term exposure levels. The present work is the result of a several week analysis of time variability of electric field level inside traffic and control channels of the GSM900 mobile communication downlink band and it indicates that a temporal model to allow future predictions of exposure on the long run is obtainable. Collecting, processing and statistically analysing the data provide expression of the maximum and weighted field strengths and their evolution in time. Specific electromagnetic footprints of the channels have been extracted, differentiations between their characteristics have been emphasised and practical advice is provided, with the scope of contributing to the development of reliable procedures for long-term exposure assessment.