Spontaneous neuronal activity prevails in virtual all brain regions in vivo as well as brain tissues in vitro. But its function, influence, or spatiotemporal patterns are ill-defined. In acute brain slice preparations, however, the level of spontaneous activity is reduced because of massively amputated neurites during slicing. Organotypic slice cultures self-restore neurites and synaptic connections and remodel the network complexity through the intrinsic rules of neural plasticity. Therefore, cultured slices could offer a unique opportunity for investigating some aspect of spontaneous activity. Here we focused on hippocampal networks and evaluated the similarity of spontaneous activity among acute slices, cultured slices and in vivo networks. Using functional multineuron calcium imaging and cell-attached patch-clamp recordings, we found that the mean firing rates of individual neurons of slice cultures did not differ from those of the in vivo hippocampus and were higher than those of acute slice neurons. Using whole-cell voltage-clamp techniques, we isolated spontaneous excitatory and inhibitory postsynaptic conductances (sEPSGs, sIPSGs, respectively) and measured five parameters, i.e., mean, coefficient of variance, skewness, kurtosis, and synaptic event frequency. The sEPSG parameters of slice cultures were similar to those of in vivo recordings, whereas the sIPSG parameters were not. We then used Soft Confidence Weighted (SCW), a machine learning technique, and tried to classify datasets. We exposed SCW to in vivo and acute slice PSG parameters of all datasets and then asked it whether the parameter sets of a given cultured slice is similar to those of either in vivo neurons or acute slices. In both EPSG and IPSG parameters, SCW judged almost all slice culture datasets as 'in vivo-like'. These results suggest that in terms of spontaneous activity, hippocampal slice cultures resemble the in vivo hippocampus, rather than acute slices.