Objective. The investigation of neurophysiologic mechanisms of anesthetic drug-induced loss of consciousness (LOC) by using the entropy, complexity, and information integration theories at the mesoscopic level has been a hot topic in recent years. However, systematic research is still lacking. Approach. We analyzed electrocorticography (ECoG) data recorded from nine rats during isoflurane-induced unconsciousness. To characterize the complexity and connectivity changes, we investigated ECoG power, symbolic dynamic-based entropy (i.e. permutation entropy (PE)), complexity (i.e. permutation Lempel-Ziv complexity (PLZC)), information integration (i.e. permutation cross mutual information (PCMI)), and PCMI-based cortical brain networks in the frontal, parietal, and occipital cortical regions. Main results. Firstly, LOC was accompanied by a raised power in the ECoG beta (12-30 Hz) but a decreased power in the high gamma (55-95 Hz) frequency band in all three brain regions. Secondly, PE and PLZC showed similar change trends in the lower frequency band (0.1-45 Hz), declining after LOC (p < 0.05) and increasing after recovery of consciousness (p < 0.001). Thirdly, intra-frontal and inter-frontal-parietal PCMI declined after LOC, in both lower (0.1-45 Hz) and higher frequency bands (55-95 Hz) (p < 0.001). Finally, the local network parameters of the nodal clustering coefficient and nodal efficiency in the frontal region decreased after LOC, in both the lower and higher frequency bands (p < 0.05). Moreover, global network parameters of the normalized average clustering coefficient and small world index increased slightly after LOC in the lower frequency band. However, this increase was not statistically significant. Significance. The PE, PLZC, PCMI and PCMI-based brain networks are effective metrics for qualifying the effects of isoflurane.