Transformers have an important role in the uninterrupted and reliable flow of electrical power in power systems. For this reason, it is necessary to monitor and control the proper and efficient operation of transformers regularly. The increase in the number of transform- ers, the differentiation of transformer operations, and the prolongation of the transformers’ operating time, accelerates the importance of remote monitoring. When remote monitor- ing structures are compared, even if conventional monitoring methods can perform many tasks, it has become suitable to establish lower cost, safer and simpler structures by us- ing the Internet of Things (IoT). Apart from monitoring the electrical parameters of the transformer, it is possible to obtain information about the operation and fault status of the transformer by monitoring the physical and chemical changes in the environment. In this study, a measurement and remote monitoring system that detects temperature, humidity, light level and gas densities are employed for low and medium voltage transformers in the indoor environment. Tests are conducted to establish a connection between the obtained ambient data and transformer operating voltages. Different classification algorithms such as Bayesian Networks (BN), Multilayer Perceptron (MLP), and Random Forest (RF) are used to classify the operating voltages versus ambient data.