Gear is one of the power transmission systems in industrial fan that is used to reduce fan rotation. Spectrum analysis can be used to detect fault in the pair of gears. However, spectrum analysis only applies to signals that are stationary and periodic. In industrial fan, the changes of fan workload causing the signals become non-stationary and non-periodic. Continuous wavelet transform (CWT) analysis is suitable for this condition. The purpose of this research is apply the CWT analysis for the gear fault identification in industrial fan. In this research, three variations of gear conditions were used (Normal, Fault level 1, and Fault level 2). Fault level 1 are pitting fault with 1.5 mm diameter in one tooth. Fault level 2 are fault with loss of one tooth. Data processing was performed by using MATLAB 2019a software. The results of data processing are grouped into 2 (original data and data after Time Synchronous Averaging (TSA)). Each variation of the gear conditions is plotting in time domain, spectrum, and CWT. The results of the research showed the CWT analysis was successfully used to identify gear fault in industrial fan. The TSA method makes the results of the CWT analysis better by reducing the noise effect. Increasing the level fault is shown by increasing value of GMF amplitude. In condition fault level 1, GMF amplitude increase 2.5 times in normal conditions. In fault level 2, the GMF amplitude increase 4 times in normal conditions and 1.5 times in fault level 1.