The representation of odor identity (ID) in the responses of olfactory neurons is not fully understood. Here, we test the hypothesis that odor ID is carried by the temporal order of activation of olfactory receptor (ORs) types. The order of OR activation is approximately preserved for different concentrations suggesting that it can carry information about the concentration-invariant odor ID. To test this hypothesis, we obtained glomerular calcium responses from the dorsal surface of the olfactory bulb to a large array of odorants, including mixtures, in multiple animals. Using this data, we computed the temporal order of glomerular activations. We proposed a neuronal distance metric in the odorant space based on rank correlation (Kendall-tau). We also measured perceptual distances in mice for the same set of odorants. We made the following observations. We find that the order-based odorant ID representations are similar across different concentrations of the same odorant. Odor IDs can be embedded into a low dimensional space (D~6). These embeddings can be accurately aligned between different animals. The representations of mixtures satisfy triangle inequality suggesting that the Kendall-tau based distances define a metric space. Using Canonical Correlation Analysis (CCA), we observed a strong correlation between the space on neural responses and the perceptual spaces of low dimension. Finally, we developed a method of embedding individual ORs into the odor space using the order-based measures. We showed that the glomeruli close to each other in the functional space can be found at similar locations in different animals, suggesting that they receive inputs from the same ORs. Overall, we developed and validated the metric for odorant representations based on the temporal sequence of OR activations. This metric yields robust representations of odor identity which generalize well across stimulus conditions and individual animals.