The study predicted the concentration of indoor total volatile organic compounds (TVOC) concentration in a randomly selected room at the Umar Kabir male Hostel located at the Federal University of Agriculture, Abeokuta, Nigeria. Readings were taken using an active sampler to measure particulate matter, PM (1.0, 2.5 and 10), TVOC, Relative Humidity (RH), Temperature and Formaldehyde. Two network types namely; feedforward back propagation and the cascaded forward back propagation were adopted randomly to predict TVOC as an output variable using data set generated from six different parameters mentioned earlier. The best performing neural network was the cascaded feed forward with a coefficient of determination of 0.98 which exhibited the lowest mean square error of 0.000124 with a network structure of 6-15-1-1. The results show the ability of Artificial Neural Networks to map inputs and outputs in complex non-linear situations such as the existence of volatile compounds in the atmosphere. It can be adopted for monitoring environmental systems by engineers and public health workers, stakeholders can use such models for initiating environmental related policies aimed at safeguarding human health.