Trace dissolved gas in transformer oil is a key feature for monitoring the transformers’ status and for estimation of early failure, therefore there is an urgent need for the development of cost-effective on line gas detection methods. As a non-destructive testing method, photoacoustic spectroscopy (PAS) technology is well-known for the advantages of high sensitivity, large dynamic range, and robustness of implementation, which has the potential for the on-line detection of trace dissolved gas detection. A T-resonator enhanced Fourier transform infrared photoacoustic spectroscopic (FTIR-PAS) system is established for the analysis of multiple trace dissolved gases in this paper. The presented T-resonator mainly consists of absorption and resonance cylinders which are perpendicular to each other. A microphone is placed on top of the resonance cylinder away from the incident light path, thus the spurious signal caused by scattered light is avoided and the noise level of the FTIR-PAS configuration is only limited by the microphone itself. The T-cell resonance frequency is determined by the dimension of the resonance cylinder, which results in a solution for the conflicting requirements of low frequency and the limited space of the FTIR compartment. The photoacoustic spectra of the mixture of 380 μL·L-1 CO2∶1 000 μL·L-1 C2H2∶N2 with 6 cm-1 resolution are collected to verify the wavenumber accuracy and multiple trace gas detection ability of the FTIR-PAS technology. The fact that the four absorption bands of CO2 and C2H2 are obviously distinguishable exhibits the good performance for simultaneous multi-dissolved gases detection of the T-resonator enhanced FTIR-PAS configuration. The results show the detection sensitivity yields 4 μL·L-1 for CO2 (light intensity 12.6 μW at 2 349 cm-1) and 5 μL·L-1 for C2H2 (30 μW at 1 360 cm-1) under STP conditions. The minimum detection limits of both gases match the national standards for transformers. By virtue of its broadband nature and high sensitivity, the T-resonator enhanced FTIR-PAS methodology is shown to be suitable for simultaneous multiple trace dissolved gas detection in transformer oil.