One of the challenges in photovoltaic solar plants is the performance maintenance in the presence of adverse environmental conditions. Soiling on the solar panels is one of those challenges having a high decrease impact on the power generation. This work proposes a statistical methodology that estimates the energy losses due to soiling on photovoltaic solar plants. Using environmental and power generation data, the proposed methodology predicts the energy generation using a regression model; and then evaluates if the differences between the observed energy generation data and the predicted energy data are due to soiling. The experiments to validate the system are based on one-year dataset of environmental and power generation data from a solar plant located in the northeast region of Brazil. The results showed that the daily energy losses estimates ranged from 2:20% up to 12:31% in a period less than a month.