It is a well-known fact that nowadays reliable traffic data are an indispensable prerequisite for efficient traffic control and planning. Nevertheless, traffic monitoring in urban areas is still difficult and expensive due to complex traffic dynamics and road networks. So, this paper describes some recent results for an innovative and cost-effective method by the author for queue length estimation at signalized intersections based on floating car data which has got the potentials to significantly enhance the quality of urban traffic monitoring. The paper starts with a short review about the new algorithm and discusses some previous observations which have been made by simulations in the virtual case of constant traffic demand. In reality, however, traffic demand typically varies over time. Often there are significant peak hours in the morning and/or in the afternoon but low traffic volumes especially at night. That is the reason why further simulations have been performed based on real daily curves of traffic volume. It turns out that the new method yields very good estimates for queue lengths at traffic signals also in this case. Even at low penetration rates regarding the available number of floating cars, the resulting daily curves of queue length are very close to the real, i.e. simulated ones. The paper discusses the results in more detail and also provides some numerical measurements for the quality of estimation. Moreover, an interesting effect is explained which sometimes occurs when floating car data of several simulation runs are aggregated. That is, surprisingly, the queue length estimates become worse in case of too much aggregation especially when general data quality and penetration rate are very good. Needless to say, this is an important phenomenon to be analyzed because it is quite usual for today’s floating car systems to combine current with historical data. It is the simplest way to get a larger amount of data points available for traffic state estimation if online penetration rates are too low for the particular applications. In summary, the paper demonstrates that the new method for queue length estimation is not only applicable to some very academic simulation examples with constant traffic volumes, but also yields highly accurate results in more realistic situations. Clearly, this is a big step forward towards future applications of the method based on real floating car data. The paper concludes with some ideas for future developments and extensions to the described algorithms and shows how the principal ideas can be used to design an integrated traffic monitoring system for urban areas. All in all, the potentials of the new method are underlined.