Several methods have been applied to the prediction of ozone concentration. In this work, an Heterogeneous Neural Network (HNN) is used to perform the same task. Different capabilities of HNN are exploited like imprecision in data or flexibility in the function computed by the neurons. The results obtained are compared with previous methodologies like Multi-layer Perceptron (MLP), Elman Network (EN), Modified Elman Network (MEN), Fuzzy Inductive Reasoning (FIR) and Long Short Term Memory Recurrent Neural Network (LSTM). Without being a Recurrent Network, HNN is able to get similar results than other methodologies, not far from them. More complex and specialized similarity functions can be developed in order to reach higher performances.