In this thesis, we step inside an unexplored region of Wireless Sensor Networks (WSNs) research. Nowadays, almost all WSNs research relies upon a hidden uniformity assumption. This assumption involves deployment, distribution and radio transmissions. Unfortunately, the real world is not uniform. In the thesis, we break the uniformity assumption and study the non-uniformity influence in WSNs. In particular, we show that addressing common WSN problems taking non-uniformity into account can provide results that are sensibly different from the ones achieved in a uniform world. In our work, we focus on the influence of non-uniformity on a particular aspect of WSNs: data management. First of all, we point out that even widely accepted solutions based on the uniformity assumption are not able to survive inside an non-uniform world. Then, we propose our approach to data management and detail a solution able to deal successfully with non-uniformity. This allows us to catch out the fundamental aspects of non-uniformity influence in WSNs and to cope with non-uniformity. Results, discussed in the thesis, show that models and solutions we propose are competitive in a uniform scenario and continue to work properly in a non-uniform world.